2013-08-26 16:28:27 +02:00
|
|
|
/*
|
2014-01-06 22:48:02 +01:00
|
|
|
* Licensed to Elasticsearch under one or more contributor
|
|
|
|
* license agreements. See the NOTICE file distributed with
|
|
|
|
* this work for additional information regarding copyright
|
|
|
|
* ownership. Elasticsearch licenses this file to you under
|
|
|
|
* the Apache License, Version 2.0 (the "License"); you may
|
|
|
|
* not use this file except in compliance with the License.
|
|
|
|
* You may obtain a copy of the License at
|
2013-08-26 16:28:27 +02:00
|
|
|
*
|
|
|
|
* http://www.apache.org/licenses/LICENSE-2.0
|
|
|
|
*
|
|
|
|
* Unless required by applicable law or agreed to in writing,
|
|
|
|
* software distributed under the License is distributed on an
|
|
|
|
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
|
|
|
|
* KIND, either express or implied. See the License for the
|
|
|
|
* specific language governing permissions and limitations
|
|
|
|
* under the License.
|
|
|
|
*/
|
|
|
|
|
2016-09-12 22:42:55 +02:00
|
|
|
package org.elasticsearch.search;
|
2013-08-26 16:28:27 +02:00
|
|
|
|
2015-04-09 18:33:27 +02:00
|
|
|
import org.apache.lucene.search.BooleanClause.Occur;
|
2015-12-18 12:43:47 -08:00
|
|
|
import org.apache.lucene.search.BooleanQuery;
|
2020-02-24 14:28:51 +00:00
|
|
|
import org.apache.lucene.search.BoostQuery;
|
2015-12-18 12:43:47 -08:00
|
|
|
import org.apache.lucene.search.Collector;
|
2016-01-12 17:40:34 +01:00
|
|
|
import org.apache.lucene.search.FieldDoc;
|
2020-01-22 11:43:15 -06:00
|
|
|
import org.apache.lucene.search.MatchNoDocsQuery;
|
2016-06-27 18:41:18 -04:00
|
|
|
import org.apache.lucene.search.Query;
|
2018-04-26 09:58:17 +02:00
|
|
|
import org.elasticsearch.Version;
|
2019-11-28 18:23:09 +01:00
|
|
|
import org.elasticsearch.action.search.SearchShardTask;
|
2013-08-26 16:28:27 +02:00
|
|
|
import org.elasticsearch.action.search.SearchType;
|
2018-04-26 09:58:17 +02:00
|
|
|
import org.elasticsearch.cluster.service.ClusterService;
|
2015-08-27 22:14:33 -04:00
|
|
|
import org.elasticsearch.common.Nullable;
|
2014-02-26 16:40:15 +01:00
|
|
|
import org.elasticsearch.common.lease.Releasables;
|
2013-08-26 16:28:27 +02:00
|
|
|
import org.elasticsearch.common.lucene.search.Queries;
|
2015-07-13 13:58:51 -04:00
|
|
|
import org.elasticsearch.common.unit.TimeValue;
|
2014-02-26 16:40:15 +01:00
|
|
|
import org.elasticsearch.common.util.BigArrays;
|
2015-04-09 18:33:27 +02:00
|
|
|
import org.elasticsearch.index.IndexService;
|
2016-01-14 15:21:59 +01:00
|
|
|
import org.elasticsearch.index.IndexSettings;
|
Upgrade master to lucene 5.0 snapshot
This has a lot of improvements in lucene, particularly around memory usage, merging, safety, compressed bitsets, etc.
On the elasticsearch side, summary of the larger changes:
API changes: postings API became a "pull" rather than "push", collector API became per-segment, etc.
packaging changes: add lucene-backwards-codecs.jar as a dependency.
improvements to boolean filtering: especially ensuring it will not be slow for SparseBitSet.
use generic BitSet api in plumbing so that concrete bitset type is an implementation detail.
use generic BitDocIdSetFilter api for dedicated bitset cache, so there is type safety.
changes to support atomic commits
implement Accountable.getChildResources (detailed memory usage API) for fielddata, etc
change handling of IndexFormatTooOld/New, since they no longer extends CorruptIndexException
Closes #8347.
Squashed commit of the following:
commit d90d53f5f21b876efc1e09cbd6d63c538a16cd89
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Nov 5 21:35:28 2014 +0100
Make default codec/postings/docvalues format constants
commit cb66c22c71cd304a36e7371b199a8c279908ae37
Merge: d4e2f6d ad4ff43
Author: Robert Muir <rmuir@apache.org>
Date: Wed Nov 5 11:41:13 2014 -0500
Merge branch 'master' into enhancement/lucene_5_0_upgrade
commit d4e2f6dfe767a5128c9b9ae9e75036378de08f47
Merge: 4e5445c 4111d93
Author: Robert Muir <rmuir@apache.org>
Date: Wed Nov 5 06:26:32 2014 -0500
Merge branch 'master' into enhancement/lucene_5_0_upgrade
commit 4e5445c775f580730eb01360244e9330c0dc3958
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 16:19:19 2014 -0500
FixedBitSet -> BitSet
commit 9887ea73e8b857eeda7f851ef3722ef580c92acf
Merge: 1bf8894 fc84666
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 15:26:25 2014 -0500
Merge branch 'master' into enhancement/lucene_5_0_upgrade
commit 1bf8894430de3e566d0dc5623b0cc28b0d674ebb
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 15:22:51 2014 -0500
remove nocommit
commit a9c2a2259ff79c69bae7806b64e92d5f472c18c8
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 13:48:43 2014 -0500
turn jenkins red again
commit 067baaaa4d52fce772c81654dcdb5051ea79139f
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 13:18:21 2014 -0500
unzip from stream
commit 82b6fba33d362aca2313cc0ca495f28f5ebb9260
Merge: b2214bb 6523cd9
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 13:10:59 2014 -0500
Merge branch 'master' into enhancement/lucene_5_0_upgrade
commit b2214bb093ec2f759003c488c3c403c8931db914
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 13:09:53 2014 -0500
go back to my URL until we can figure out what is up with jenkins
commit e7d614172240175a51f580aeaefb6460d21cede9
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 10:52:54 2014 -0500
try this jenkins
commit 337a3c7704efa7c9809bf373152d711ee55f876c
Author: Simon Willnauer <simonw@apache.org>
Date: Tue Nov 4 16:17:49 2014 +0100
Rename temp-files under lock to prevent metadata reads while renaming
commit 77d5ba80d0a76efa549dd753b9f114b2f2d2d29c
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 10:07:11 2014 -0500
continue to treat too-old/too-new as corruption for now
commit 98d0fd2f4851bc50e505a94ca592a694d502c51c
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 09:24:21 2014 -0500
fix last nocommit
commit 643fceed66c8caf22b97fc489d67b4a2a90a1a1c
Author: Simon Willnauer <simonw@apache.org>
Date: Tue Nov 4 14:46:17 2014 +0100
remove NoSuchDirectoryException
commit 2e43c4feba05cfaf451df70f946c0930cbcc4557
Merge: 93826e4 8163107
Author: Simon Willnauer <simonw@apache.org>
Date: Tue Nov 4 14:38:00 2014 +0100
Merge branch 'master' into enhancement/lucene_5_0_upgrade
commit 93826e4d56a6a97c2074669014af77ff519bde63
Merge: 7f10129 44e24d3
Author: Simon Willnauer <simonw@apache.org>
Date: Tue Nov 4 12:54:27 2014 +0100
Merge branch 'master' into enhancement/lucene_5_0_upgrade
Conflicts:
src/main/java/org/elasticsearch/index/store/DistributorDirectory.java
src/main/java/org/elasticsearch/index/store/Store.java
src/main/java/org/elasticsearch/indices/recovery/RecoveryStatus.java
src/test/java/org/elasticsearch/index/store/DistributorDirectoryTest.java
src/test/java/org/elasticsearch/index/store/StoreTest.java
src/test/java/org/elasticsearch/indices/recovery/RecoveryStatusTests.java
commit 7f10129364623620575c109df725cf54488b3abb
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Nov 4 11:32:24 2014 +0100
Fix TopHitsAggregator to not ignore the top-level/leaf collector split.
commit 042fadc8603b997bdfdc45ca44fec70dc86774a6
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Nov 4 11:31:20 2014 +0100
Remove MatchDocIdSet in favor of DocValuesDocIdSet.
commit 7d877581ff5db585a674c95ac391ac78a0282826
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Nov 4 11:10:08 2014 +0100
Make the and filter use the cost API.
Lucene 5 ensured that cost() can safely be used, and this will have the benefit
that the order in which filters are specified is not important anymore (only
for slow random-access filters in practice).
commit 78f1718aa2cd82184db7c3a8393e6215f43eb4a8
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 23:55:17 2014 -0500
fix previous eclipse import braindamage
commit 186c40e9258ce32f22a9a714ab442a310b6376e0
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 22:32:34 2014 -0500
allow child queries to exhaust iterators again
commit b0b1271305e1b6d0c4c4da51a3c54df1aa5c0605
Author: Ryan Ernst <ryan@iernst.net>
Date: Mon Nov 3 14:50:44 2014 -0800
Fix nocommit for mapping output. index_options will not be printed if
the field is not indexed.
commit ba223eb85e399c9620a347a983e29bf703953e7a
Author: Ryan Ernst <ryan@iernst.net>
Date: Mon Nov 3 14:07:26 2014 -0800
Remove no commit for chinese analyzer provider. We should have a
separate issue to address not using this provider on new indexes.
commit ca554b03c4471797682b2fb724f25205cf040c4a
Author: Ryan Ernst <ryan@iernst.net>
Date: Mon Nov 3 13:41:59 2014 -0800
Fix stop tests
commit de67c4653ec47dee9c671390536110749d2bb05f
Author: Ryan Ernst <ryan@iernst.net>
Date: Mon Nov 3 12:51:17 2014 -0800
Remove analysis nocommits, switching over to Lucene43*Filters for
backcompat
commit 50cae9bec72c25c33a1ab8a8931bccb3355171e2
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 15:32:25 2014 -0500
add ram accounting and TODO lazy-loading (its no worse than master, can be a followup improvement) for suggesters
commit 7a7f0122f138684b312d0f0b03dc2a9c16c15f9c
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 15:11:26 2014 -0500
bump lucene version
commit cd0cae5c35e7a9e049f49ae45431f658fb86676b
Merge: 446bc09 3c72073
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 14:49:05 2014 -0500
Merge branch 'master' into enhancement/lucene_5_0_upgrade
commit 446bc09b4e8bf4602d3c252b53ddaa0da65cce2f
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 14:46:30 2014 -0500
remove hack
commit a19d85a968d82e6d00292b49630ef6ff2dbf2f32
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 12:53:11 2014 -0500
dont create exceptions with circular references on corruption (will open a PR for this)
commit 0beefb9e821d97c37e90ec556d81ac7b00369b8a
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 11:47:14 2014 -0500
temporarily add craptastic detector for this horrible bug
commit e9f2d298bff75f3d1591f8622441e459c3ce7ac3
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 10:56:01 2014 -0500
add nocommit
commit e97f1d50a91a7129650b8effc7a9ecf74ca0569a
Merge: c57a3c8 f1f50ac
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 10:12:12 2014 -0500
Merge branch 'master' into enhancement/lucene_5_0_upgrade
commit c57a3c8341ed61dca62eaf77fad6b8b48aeb6940
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 10:11:46 2014 -0500
fix nocommit
commit dd0e77e4ec07c7011ab5f6b60b2ead33dc2333d2
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 09:54:09 2014 -0500
nocommit -> TODO, this is in much more places in the codebase, bigger issue
commit 3cc3bf56d72d642059f8fe220d6f2fed608363e9
Author: Ryan Ernst <ryan@iernst.net>
Date: Sat Nov 1 23:59:17 2014 -0700
Remove nocommit and awaitsfix for edge ngram filter test.
commit 89f115245155511c0fbc0d5ee62e63141c3700c1
Author: Ryan Ernst <ryan@iernst.net>
Date: Sat Nov 1 23:57:44 2014 -0700
Fix EdgeNGramTokenFilter logic for version <= 4.3, and fixed instanceof
checks in corresponding tests to correctly check for reverse filter when
applicable.
commit 112df869cd199e36aab0e1a7a288bb1fdb2ebf1c
Author: Robert Muir <rmuir@apache.org>
Date: Sun Nov 2 00:08:30 2014 -0400
execute geo disjoint query/filter as intersects
commit e5061273cc685f1252e9a3a9ae4877ec9bce7752
Author: Robert Muir <rmuir@apache.org>
Date: Sat Nov 1 22:58:59 2014 -0400
remove chinese analyzer from docs
commit ea1af11b8978fcc551f198e24fe21d52806993ef
Author: Robert Muir <rmuir@apache.org>
Date: Sat Nov 1 22:29:00 2014 -0400
fix ram accounting bug
commit 53c0a42c6aa81aa6bf81d3aa77b95efd513e0f81
Merge: e3bcd3c 6011a18
Author: Robert Muir <rmuir@apache.org>
Date: Sat Nov 1 22:16:29 2014 -0400
Merge branch 'master' into enhancement/lucene_5_0_upgrade
commit e3bcd3cc07a4957e12c7b3affc462c31290a9186
Author: Robert Muir <rmuir@apache.org>
Date: Sat Nov 1 22:15:01 2014 -0400
fix url-email back compat (thanks ryan)
commit 91d6b096a96c357755abee167098607223be1aad
Author: Robert Muir <rmuir@apache.org>
Date: Sat Nov 1 22:11:26 2014 -0400
bump lucene version
commit d2bb9568df72b37ec7050d25940160b8517394bc
Author: Robert Muir <rmuir@apache.org>
Date: Sat Nov 1 20:33:07 2014 -0400
remove nocommit
commit 1d049c471e19e5c457262c7399c5bad9e023b2e3
Author: Robert Muir <rmuir@apache.org>
Date: Sat Nov 1 20:28:58 2014 -0400
fix eclipse to group org/com imports together: without this, its madness
commit 09d8c1585ee99b6e63be032732c04ef6fed84ed2
Author: Robert Muir <rmuir@apache.org>
Date: Sat Nov 1 14:27:41 2014 -0400
remove nocommit, if you dont liek it, print assembly and tell me how it can be better
commit 8a6a294313fdf33b50c7126ec20c07867ecd637c
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 31 20:01:55 2014 +0100
Remove deprecated usage of DocIdSets.newDocIDSet.
commit 601bee60543610558403298124a84b1b3bbd1045
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 31 14:13:18 2014 -0400
maybe one of these zillions of annotations will stop thread leaks
commit 9d3f69abc7267c5e455aefa26db95cb554b02d62
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 31 14:05:39 2014 -0400
fix some analysis nocommits
commit 312e3a29c77214b8142d21c33a6b2c2b151acf9a
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 31 18:28:45 2014 +0100
Remove XConstantScoreQuery/XFilteredQuery/ApplyAcceptedDocsFilter.
commit 5a0cb9f8e167215df7f1b1fad11eec6e6c74940f
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 31 17:06:45 2014 +0100
Fix misleading documentation of DocIdSets.toCacheable.
commit 8b4ef2b5b476fff4c79c0c2a0e4769ead26cf82b
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 31 17:05:59 2014 +0100
Fix CustomRandomAccessFilterStrategy to override the right method.
commit d7a9a407a615987cfffc651f724fbd8795c9c671
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 31 16:21:35 2014 +0100
Better handle the special case when there is a single SHOULD clause.
commit 648ad389f07e92dfc451f345549c9841ba5e4c9a
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 31 15:53:38 2014 +0100
Cut over XBooleanFilter to BitDocIdSet.Builder.
The idea is similar to what happened to Lucene's BooleanFilter.
Yet XBooleanFilter is a bit more sophisticated and I had to slightly
change the way it is implemented in order to make it work. The main difference
with before is that slow filters are now applied lazily, so eg. if you have 3
MUST clauses, two with a fast iterator and the third with a slow iterator, the
previous implementation used to apply the fast iterators first and then only
check the slow filter for bits which were set in the bit set. Now we are
computing a bit set based on the fast must clauses and then basically returning
a BitsFilteredDocIdSet.wrap(bitset, slowClause).
Other than that, BooleanFilter still uses the bitset optimizations when or-ing
and and-ind filters.
Another improvement is that BooleanFilter is now aware of the cost API.
commit b2dad312b4bc9f931dc3a25415dd81c0d9deee08
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 31 10:18:53 2014 -0400
clear nocommit
commit 4851d2091e744294336dfade33906c75fbe695cd
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 15:15:16 2014 +0100
cut over to RoaringDocIdSet
commit ca6aec24a901073e65ce4dd6b70964fd3612409e
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 14:57:30 2014 +0100
make nocommit more explicit
commit d0742ee2cb7a6c48b0bbb31580b7fbcebdb6ec40
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 31 09:55:24 2014 -0400
fix standardtokenizer nocommit
commit 7d6faccafff22a86af62af0384838391d46695ca
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 14:54:08 2014 +0100
fix compilation
commit a038a405c1ff6458ad294e6b5bc469e622f699d0
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 14:53:43 2014 +0100
fix compilation
commit 30c9e307b1f5d80e2deca3392c0298682241207f
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 14:52:35 2014 +0100
fix compilation
commit e5139bc5a0a9abd2bdc6ba0dfbcb7e3c2e7b8481
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 31 09:52:16 2014 -0400
clear nocommit here
commit 85dd2cedf7a7994bed871ac421cfda06aaf5c0a5
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 14:46:17 2014 +0100
fix CompletionPostingsFormatTest
commit c0f3781f616c9b0ee3b5c4d0998810f595868649
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 31 09:38:00 2014 -0400
add tests for these analyzers
commit 51f9999b4ad079c283ae762c862fd0e22d00445f
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 14:10:26 2014 +0100
remove nocommit - this is not an issue
commit fd1388fa03e622b0738601c8aeb2dbf7949a6dd2
Author: Martijn van Groningen <martijn.v.groningen@gmail.com>
Date: Fri Oct 31 14:07:01 2014 +0100
Remove redundant null check
commit 3d6dd51b0927337ba941a235446b22e8cd500dc3
Author: Martijn van Groningen <martijn.v.groningen@gmail.com>
Date: Fri Oct 31 14:01:37 2014 +0100
Removed the work around to prevent p/c error when invoking #iterator() twice, because the custom query filter wrapper now doesn't transform the result to a cache doc id set any more.
I think the transforming to a cachable doc id set in CustomQueryWrappingFilter isn't needed at all, because we use the DocIdSet only once and because of that is just slowed things down.
commit 821832a537e00cd1216064b379df3e01d2911d3a
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 13:54:33 2014 +0100
one more nocommit
commit 77eb9ea4c4ea50afb2680c29682ddcb3851a9d4f
Author: Martijn van Groningen <martijn.v.groningen@gmail.com>
Date: Fri Oct 31 13:52:29 2014 +0100
Remove cast
commit a400573c034ed602221f801b20a58a9186a06eae
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 13:49:24 2014 +0100
fix stop filter
commit 51746087cf8ec34c4d20aa05ba8dbff7b3b43eec
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 13:21:36 2014 +0100
fix changed semantics of FBS.nextSetBit to check for NO_MORE_DOCS
commit 8d0a4e2511310f1293860823fe3ba80ac771bbe3
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 31 08:13:44 2014 -0400
do the bogus cast differently
commit 46a5cc5732dea096c0c80ae5ce42911c9c51e44e
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 13:00:16 2014 +0100
I hate it but P/C now passes
commit 580c0c2f82bbeacf217e594f22312b11d1bdb839
Merge: a9d3c00 1645434
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 31 06:54:31 2014 -0400
fix nocommit/classcast
commit a9d3c004d62fe04989f49a897e6ff84973c06eb9
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 31 08:49:31 2014 +0100
Update TODO.
commit aa75af0b407792aeef32017f03a6f442ed970baa
Author: Robert Muir <rmuir@apache.org>
Date: Thu Oct 30 19:18:25 2014 -0400
clear obselete nocommits from lucene bump
commit d438534cf41fcbe2d88070e2f27c994625e082c2
Author: Robert Muir <rmuir@apache.org>
Date: Thu Oct 30 18:53:20 2014 -0400
throw classcastexception when ES abuses regular filtercache for nested docs
commit 2c751f3a8feda43ec127c34769b069de21f3d16f
Author: Robert Muir <rmuir@apache.org>
Date: Thu Oct 30 18:31:34 2014 -0400
bump lucene revision, fix tests
commit d6ef7f6304ae262bf6228a7d661b2a452df332be
Author: Simon Willnauer <simonw@apache.org>
Date: Thu Oct 30 22:37:58 2014 +0100
fix merge problems
commit de9d361f88a9ce6bb3fba85285de41f223c95767
Merge: 41f6aab f6b37a3
Author: Simon Willnauer <simonw@apache.org>
Date: Thu Oct 30 22:28:59 2014 +0100
Merge branch 'master' into enhancement/lucene_5_0_upgrade
Conflicts:
pom.xml
src/main/java/org/elasticsearch/Version.java
src/main/java/org/elasticsearch/gateway/local/state/meta/MetaDataStateFormat.java
commit 41f6aab388aa80c40b08a2facab2617576203a0d
Author: Simon Willnauer <simonw@apache.org>
Date: Thu Oct 30 17:48:46 2014 +0100
fix potiential NPE
commit c4428b12e1ae838b91e847df8b4a8be7f49e10f4
Author: Simon Willnauer <simonw@apache.org>
Date: Thu Oct 30 17:38:46 2014 +0100
don't advance iterator in a match(doc) method
commit 28ab948e99e3ea4497c9b1e468384806ba7e1790
Author: Simon Willnauer <simonw@apache.org>
Date: Thu Oct 30 17:34:58 2014 +0100
don't advance iterator in a match(doc) method
commit eb0f33f6634fadfcf4b2bf7327400e568f0427bb
Author: Simon Willnauer <simonw@apache.org>
Date: Thu Oct 30 16:55:54 2014 +0100
fix GeoUtilsTest
commit 7f711fe3eaf73b6c2268cf42d5a41132a61ad831
Author: Simon Willnauer <simonw@apache.org>
Date: Thu Oct 30 16:43:16 2014 +0100
Use a dedicated default index option if field type is not indexed by default
commit 78e3f37ab779e3e1b25b45a742cc86ab5f975149
Author: Robert Muir <rmuir@apache.org>
Date: Thu Oct 30 10:56:14 2014 -0400
disable this test with AwaitsFix to reduce noise
commit 9a590f563c8e03a99ecf0505c92d12d7ab20d11d
Author: Simon Willnauer <simonw@apache.org>
Date: Thu Oct 30 09:38:49 2014 +0100
fix lucene version
commit abe3ca1d8bb6b5101b545198f59aec44bacfa741
Author: Simon Willnauer <simonw@apache.org>
Date: Thu Oct 30 09:35:05 2014 +0100
fix AnalyzingCompletionLookupProvider to wrok with new codec API
commit 464293b245852d60bde050c6d3feb5907dcfbf5f
Author: Robert Muir <rmuir@apache.org>
Date: Thu Oct 30 00:26:00 2014 -0400
don't try to write stuff to tests class directory
commit 031cc6c19f4fe4423a034b515f77e5a0e282a124
Author: Robert Muir <rmuir@apache.org>
Date: Thu Oct 30 00:12:36 2014 -0400
AwaitsFix these known issues to reduce noise
commit 4600d51891e35847f2d344247d6f915a0605c0d1
Author: Robert Muir <rmuir@apache.org>
Date: Thu Oct 30 00:06:53 2014 -0400
openbitset lives on
commit 8492bae056249e2555d24acd55f1046b66a667c4
Author: Robert Muir <rmuir@apache.org>
Date: Wed Oct 29 23:42:54 2014 -0400
fixes for filter tests
commit 31f24ce4efeda31f97eafdb122346c7047a53bf2
Author: Robert Muir <rmuir@apache.org>
Date: Wed Oct 29 23:12:38 2014 -0400
don't use fieldcache
commit 8480789942fdff14a6d2b2cd8134502fe62f20c8
Author: Robert Muir <rmuir@apache.org>
Date: Wed Oct 29 23:04:29 2014 -0400
ancient index no longer supported
commit 02e78dc7ebdd827533009f542582e8db44309c57
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 23:37:02 2014 +0100
fix more tests
commit ff746c6df23c50b3f3ec24922413b962c8983080
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 23:08:19 2014 +0100
fix all mapper
commit e4fb84b517107b25cb064c66f83c9aa814a311b2
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 22:55:54 2014 +0100
fix distributor tests and cut over to FileStore API
commit 20c850e2cfe3210cd1fb9e232afed8d4ac045857
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 22:42:18 2014 +0100
use DOCS_ONLY if index=true and current options == null
commit 44169c108418413cfe51f5ce23ab82047463e4c2
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 22:33:36 2014 +0100
Fix index=yes|no settings in mappers
commit a3c5f77987461a18121156ed345d42ded301c566
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 21:51:41 2014 +0100
fix several field mappers conversion from setIndexed to indexOptions
commit df84d736908e88a031d710f98e222be68ae96af1
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 21:33:35 2014 +0100
fix SourceFieldMapper to be not indexed
commit b2bf01d12a8271a31fb2df601162d0e89924c8f5
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 21:23:08 2014 +0100
Cut over to .liv files in store and corruption tests
commit 619004df436f9ef05d24bef1b6a7f084c6b0ad75
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 17:05:52 2014 +0100
fix more tests
commit b7ed653a8b464de446e00456bce0a89e47627c38
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 16:19:08 2014 +0100
[STORE] Add dedicated method to write temporary files
Recovery writes temporary files which might not end up in the
right distributor directories today. This commit adds a dedicated
API that allows specifying the target file name in order to create the
tempoary file in the correct directory.
commit 7d574659f6ae04adc2b857146ad0d8d56ca66f12
Author: Robert Muir <rmuir@apache.org>
Date: Wed Oct 29 10:28:49 2014 -0400
add some leniency to temporary bogus method
commit f97022ea7c2259f7a5cf97d924c59ed75ab65b32
Author: Robert Muir <rmuir@apache.org>
Date: Wed Oct 29 10:24:17 2014 -0400
fix MultiCollector bug
commit b760533128c2b4eb10ad76e9689ef714293dd819
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 14:56:08 2014 +0100
CheckIndex is now closeable we need to close it
commit 9dae9fb6d63546a6c2427be2a2d5c8358f5b1934
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 14:45:11 2014 +0100
s/Lucene51/Lucene50
commit 7aea9b86856a8c1b06a08e7c312ede1168af1287
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 14:42:30 2014 +0100
fix BloomFilterPostingsFormat
commit 16fea6fe842e88665d59cc091e8224e8dc6ce08c
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 14:41:16 2014 +0100
fix some codec format issues
commit 3d77aa97dd2c4012b63befef3f2ba2525965e8a6
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 14:30:43 2014 +0100
fix CodecTests
commit 6ef823b1fde25657438ace1aabd9d552d6ae215e
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 14:26:47 2014 +0100
make it compile
commit 9991eee1fe99435118d4dd42b297ffc83fce5ec5
Author: Robert Muir <rmuir@apache.org>
Date: Wed Oct 29 09:12:43 2014 -0400
add an ugly hack for TopHitsAggregator for now
commit 03e768a01fcae6b1f4cb50bcceec7d42977ac3e6
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 14:01:02 2014 +0100
cut over ES090PostingsFormat
commit 463d281faadb794fdde3b469326bdaada25af048
Merge: 0f8740a 8eac79c
Author: Robert Muir <rmuir@apache.org>
Date: Wed Oct 29 08:30:36 2014 -0400
Merge branch 'master' into enhancement/lucene_5_0_upgrade
commit 0f8740a782455a63524a5a82169f6bbbfc613518
Author: Robert Muir <rmuir@apache.org>
Date: Wed Oct 29 01:00:15 2014 -0400
fix/hack remaining filter and analysis issues
commit df534488569da13b31d66e581456dfd4b55156b9
Author: Robert Muir <rmuir@apache.org>
Date: Tue Oct 28 23:11:47 2014 -0400
fix ngrams / openbitset usage
commit 11f5dc3b9887f4da80a0fa1818e1350b30599329
Author: Robert Muir <rmuir@apache.org>
Date: Tue Oct 28 22:42:44 2014 -0400
hack over sort comparators
commit 4ebdc754350f512596f6a02770d223e9f5f7975a
Author: Robert Muir <rmuir@apache.org>
Date: Tue Oct 28 21:27:07 2014 -0400
compiler errors < 100
commit 2d60c9e29de48ccb0347dd87f7201f47b67b83a0
Author: Robert Muir <rmuir@apache.org>
Date: Tue Oct 28 03:13:08 2014 -0400
clear some nocommits around ram usage
commit aaf47fe6c0aabcfb2581dd456fc50edf871da758
Author: Robert Muir <rmuir@apache.org>
Date: Mon Oct 27 12:27:34 2014 -0400
migrate fieldinfo handling
commit ef6ed6d15d8def71cd880d97249678136cd29fe3
Author: Robert Muir <rmuir@apache.org>
Date: Mon Oct 27 12:07:13 2014 -0400
more simple fixes
commit f475e1048ae697dd9da5bd9da445102b0b7bc5b3
Author: Robert Muir <rmuir@apache.org>
Date: Mon Oct 27 11:58:21 2014 -0400
more fielddata ram accounting fixes
commit 16b4239eaa9b4262df258257df4f31d39f28a3a2
Author: Simon Willnauer <simonw@apache.org>
Date: Mon Oct 27 16:47:32 2014 +0100
add missing file
commit 5b542fa2a6da81e36a0c35b8e891a1d8bc58f663
Author: Simon Willnauer <simonw@apache.org>
Date: Mon Oct 27 16:43:29 2014 +0100
cut over completion posting formats - still some nocommits
commit ecdea49404c4ec4e1b78fb54575825f21b4e096e
Author: Robert Muir <rmuir@apache.org>
Date: Mon Oct 27 11:21:09 2014 -0400
fielddata accountable fixes
commit d43da265718917e20c8264abd43342069198fe9c
Author: Simon Willnauer <simonw@apache.org>
Date: Mon Oct 27 16:19:53 2014 +0100
cut over BloomFilterPostings to new API
commit 29b192ba621c14820175775d01242162b88bd364
Author: Robert Muir <rmuir@apache.org>
Date: Mon Oct 27 10:22:51 2014 -0400
fix more analyzers
commit 74b4a0c5283e323a7d02490df469497c722780d2
Author: Robert Muir <rmuir@apache.org>
Date: Mon Oct 27 09:54:25 2014 -0400
fix tests
commit 554084ccb4779dd6b1c65fa7212ad1f64f3a6968
Author: Simon Willnauer <simonw@apache.org>
Date: Mon Oct 27 14:51:48 2014 +0100
maintain supressed exceptions on CorruptIndexException
commit cf882d9112c5e8ef1e9f2b0f800f7aa59001a4f2
Author: Simon Willnauer <simonw@apache.org>
Date: Mon Oct 27 14:47:17 2014 +0100
commitOnClose=false
commit ebb2a9189ab2f459b7c6c9985be610fd90dfe410
Author: Simon Willnauer <simonw@apache.org>
Date: Mon Oct 27 14:46:06 2014 +0100
cut over indexwriter closeing in InternalEngine
commit cd21b3d4706f0b562bd37792d077d60832aff65f
Author: Simon Willnauer <simonw@apache.org>
Date: Mon Oct 27 14:38:10 2014 +0100
fix constant
commit f93f900c4a1c90af3a21a4af5735a7536423fe28
Author: Robert Muir <rmuir@apache.org>
Date: Mon Oct 27 09:50:49 2014 -0400
fix test
commit a9a752940b1ab4699a6a08ba8b34afca82b843fe
Author: Martijn van Groningen <martijn.v.groningen@gmail.com>
Date: Mon Oct 27 09:26:18 2014 +0100
Be explicit about the index options
commit d9ee815babd030fa2ceaec9f467c105ee755bf6b
Author: Simon Willnauer <simonw@apache.org>
Date: Sun Oct 26 20:03:44 2014 +0100
cut over store and directory
commit b3f5c8e39039dd8f5caac0c4dd1fc3b1116e64ca
Author: Robert Muir <rmuir@apache.org>
Date: Sun Oct 26 13:08:39 2014 -0400
more test fixes
commit 8842f2684e3606aae0860c27f7a4c53e273d47fb
Author: Robert Muir <rmuir@apache.org>
Date: Sun Oct 26 12:14:52 2014 -0400
tests manual labor
commit c43de5aec337919a3fdc3638406dff17fc80bc98
Author: Robert Muir <rmuir@apache.org>
Date: Sun Oct 26 11:04:13 2014 -0400
BytesRef -> BytesRefBuilder
commit 020c0d087a2f37566a1db390b0e044ebab030138
Author: Martijn van Groningen <martijn.v.groningen@gmail.com>
Date: Sun Oct 26 15:53:37 2014 +0100
Moved over to BitSetFilter
commit 48dd1b909e6c52cef733961c9ecebfe4f67109fe
Author: Martijn van Groningen <martijn.v.groningen@gmail.com>
Date: Sun Oct 26 15:53:11 2014 +0100
Left over Collector api change in ScanContext
commit 6ec248ef63f262bcda400181b838fd9244752625
Author: Martijn van Groningen <martijn.v.groningen@gmail.com>
Date: Sun Oct 26 15:47:40 2014 +0100
Moved indexed() over to indexOptions != null or indexOptions == null
commit 9937aebfd8546ae4bb652cd976b3b43ac5ab7a63
Author: Martijn van Groningen <martijn.v.groningen@gmail.com>
Date: Sun Oct 26 13:26:31 2014 +0100
Fixed many compile errors. Mainly around the breaking Collector api change in 5.0.
commit fec32c4abc0e3309cf34260c8816305a6f820c9e
Author: Robert Muir <rmuir@apache.org>
Date: Sat Oct 25 11:22:17 2014 -0400
more easy fixes
commit dab22531d801800d17a65dc7c9464148ce8ebffd
Author: Robert Muir <rmuir@apache.org>
Date: Sat Oct 25 09:33:41 2014 -0400
more progress
commit 414767e9a955010076b0497cc4f6d0c1850b48d3
Author: Robert Muir <rmuir@apache.org>
Date: Sat Oct 25 06:33:17 2014 -0400
more progress
commit ad9d969fddf139a8830254d3eb36a908ba87cc12
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 24 14:28:01 2014 -0400
current state of fun
commit 464475eecb0be15d7d084135ed16051f76a7e521
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 24 11:42:41 2014 -0400
bump to 5.0 snapshot
2014-11-05 15:48:51 -05:00
|
|
|
import org.elasticsearch.index.cache.bitset.BitsetFilterCache;
|
2014-09-04 02:54:19 +02:00
|
|
|
import org.elasticsearch.index.engine.Engine;
|
2017-07-26 09:16:52 +02:00
|
|
|
import org.elasticsearch.index.fielddata.IndexFieldData;
|
2015-06-09 09:56:39 -07:00
|
|
|
import org.elasticsearch.index.mapper.MappedFieldType;
|
2013-08-26 16:28:27 +02:00
|
|
|
import org.elasticsearch.index.mapper.MapperService;
|
2016-08-10 16:19:21 +02:00
|
|
|
import org.elasticsearch.index.mapper.ObjectMapper;
|
|
|
|
import org.elasticsearch.index.mapper.TypeFieldMapper;
|
2015-10-23 11:55:28 +02:00
|
|
|
import org.elasticsearch.index.query.AbstractQueryBuilder;
|
2013-08-26 16:28:27 +02:00
|
|
|
import org.elasticsearch.index.query.ParsedQuery;
|
2016-10-14 16:26:09 +02:00
|
|
|
import org.elasticsearch.index.query.QueryBuilder;
|
2016-02-08 12:09:15 +01:00
|
|
|
import org.elasticsearch.index.query.QueryShardContext;
|
Nested queries should avoid adding unnecessary filters when possible. (#23079)
When nested objects are present in the mappings, many queries get deoptimized
due to the need to exclude documents that are not in the right space. For
instance, a filter is applied to all queries that prevents them from matching
non-root documents (`+*:* -_type:__*`). Moreover, a filter is applied to all
child queries of `nested` queries in order to make sure that the child query
only matches child documents (`_type:__nested_path`), which is required by
`ToParentBlockJoinQuery` (the Lucene query behing Elasticsearch's `nested`
queries).
These additional filters slow down `nested` queries. In 1.7-, the cost was
somehow amortized by the fact that we cached filters very aggressively. However,
this has proven to be a significant source of slow downs since 2.0 for users
of `nested` mappings and queries, see #20797.
This change makes the filtering a bit smarter. For instance if the query is a
`match_all` query, then we need to exclude nested docs. However, if the query
is `foo: bar` then it may only match root documents since `foo` is a top-level
field, so no additional filtering is required.
Another improvement is to use a `FILTER` clause on all types rather than a
`MUST_NOT` clause on all nested paths when possible since `FILTER` clauses
are more efficient.
Here are some examples of queries and how they get rewritten:
```
"match_all": {}
```
This query gets rewritten to `ConstantScore(+*:* -_type:__*)` on master and
`ConstantScore(_type:AutomatonQuery {\norg.apache.lucene.util.automaton.Automaton@4371da44})`
with this change. The automaton is the complement of `_type:__*` so it matches
the same documents, but is faster since it is now a positive clause. Simplistic
performance testing on a 10M index where each root document has 5 nested
documents on average gave a latency of 420ms on master and 90ms with this change
applied.
```
"term": {
"foo": {
"value": "0"
}
}
```
This query is rewritten to `+foo:0 #(ConstantScore(+*:* -_type:__*))^0.0` on
master and `foo:0` with this change: we do not need to filter nested docs out
since the query cannot match nested docs. While doing performance testing in
the same conditions as above, response times went from 250ms to 50ms.
```
"nested": {
"path": "nested",
"query": {
"term": {
"nested.foo": {
"value": "0"
}
}
}
}
```
This query is rewritten to
`+ToParentBlockJoinQuery (+nested.foo:0 #_type:__nested) #(ConstantScore(+*:* -_type:__*))^0.0`
on master and `ToParentBlockJoinQuery (nested.foo:0)` with this change. The
top-level filter (`-_type:__*`) could be removed since `nested` queries only
match documents of the parent space, as well as the child filter
(`#_type:__nested`) since the child query may only match nested docs since the
`nested` object has both `include_in_parent` and `include_in_root` set to
`false`. While doing performance testing in the same conditions as above,
response times went from 850ms to 270ms.
2017-02-14 16:05:19 +01:00
|
|
|
import org.elasticsearch.index.search.NestedHelper;
|
2014-12-11 16:38:17 +01:00
|
|
|
import org.elasticsearch.index.shard.IndexShard;
|
2013-08-26 16:28:27 +02:00
|
|
|
import org.elasticsearch.index.similarity.SimilarityService;
|
2013-11-24 03:13:08 -08:00
|
|
|
import org.elasticsearch.search.aggregations.SearchContextAggregations;
|
2017-01-23 16:33:51 +01:00
|
|
|
import org.elasticsearch.search.collapse.CollapseContext;
|
2013-08-26 16:28:27 +02:00
|
|
|
import org.elasticsearch.search.dfs.DfsSearchResult;
|
2015-12-16 10:22:54 +00:00
|
|
|
import org.elasticsearch.search.fetch.FetchPhase;
|
2013-08-26 16:28:27 +02:00
|
|
|
import org.elasticsearch.search.fetch.FetchSearchResult;
|
2016-09-07 23:43:17 +02:00
|
|
|
import org.elasticsearch.search.fetch.StoredFieldsContext;
|
2020-02-13 12:59:59 -08:00
|
|
|
import org.elasticsearch.search.fetch.subphase.FetchDocValuesContext;
|
2020-07-28 10:58:20 -07:00
|
|
|
import org.elasticsearch.search.fetch.subphase.FetchFieldsContext;
|
2016-07-29 23:05:38 -04:00
|
|
|
import org.elasticsearch.search.fetch.subphase.FetchSourceContext;
|
|
|
|
import org.elasticsearch.search.fetch.subphase.ScriptFieldsContext;
|
2020-07-29 11:39:17 -07:00
|
|
|
import org.elasticsearch.search.fetch.subphase.highlight.SearchHighlightContext;
|
2016-09-12 22:42:55 +02:00
|
|
|
import org.elasticsearch.search.internal.ContextIndexSearcher;
|
|
|
|
import org.elasticsearch.search.internal.ScrollContext;
|
|
|
|
import org.elasticsearch.search.internal.SearchContext;
|
2020-03-09 11:59:30 -04:00
|
|
|
import org.elasticsearch.search.internal.SearchContextId;
|
2016-09-12 22:42:55 +02:00
|
|
|
import org.elasticsearch.search.internal.ShardSearchRequest;
|
Add ability to profile query and collectors
Provides a new flag which can be enabled on a per-request basis.
When `"profile": true` is set, the search request will execute in a
mode that collects detailed timing information for query components.
```
GET /test/test/_search
{
"profile": true,
"query": {
"match": {
"foo": "bar"
}
}
}
```
Closes #14889
Squashed commit of the following:
commit a92db5723d2c61b8449bd163d2f006d12f9889ad
Merge: 117dd99 3f87b08
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Dec 17 09:44:10 2015 -0500
Merge remote-tracking branch 'upstream/master' into query_profiler
commit 117dd9992e8014b70203c6110925643769d80e62
Merge: 9b29d68 82a64fd
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Dec 15 13:27:18 2015 -0500
Merge remote-tracking branch 'upstream/master' into query_profiler
Conflicts:
core/src/main/java/org/elasticsearch/search/SearchService.java
commit 9b29d6823a71140ecd872df25ff9f7478e7fe766
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Dec 14 16:13:23 2015 -0500
[TEST] Profile flag needs to be set, ensure searches go against primary only for consistency
commit 4d602d8ad1f8cbc7b475450921fa3bc7d395b34f
Merge: 8b48e87 7742c1e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Dec 14 10:56:25 2015 -0500
Merge remote-tracking branch 'upstream/master' into query_profiler
commit 8b48e876348b163ab730eeca7fa35142165b05f9
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Dec 14 10:56:01 2015 -0500
Delegate straight to in.matchCost, no need for profiling
commit fde3b0587911f0b5f15e779c671d0510cbd568a9
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Dec 14 10:28:23 2015 -0500
Documentation tweaks, renaming build_weight -> create_weight
commit 46f5e011ee23fe9bb8a1f11ceb4fa9d19fe48e2e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Dec 14 10:27:52 2015 -0500
Profile TwoPhaseIterator should override matchCost()
commit b59f894ddb11b2a7beebba06c4ec5583ff91a7b2
Merge: 9aa1a3a b4e0c87
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 9 14:23:26 2015 -0500
Merge remote-tracking branch 'upstream/master' into query_profiler
commit 9aa1a3a25c34c9cd9fffaa6114c25a0ec791307d
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 9 13:41:48 2015 -0500
Revert "Move some of the collector wrapping logic into ProfileCollectorBuilder"
This reverts commit 02cc31767fb76a7ecd44a302435e93a05fb4220e.
commit 57f7c04cea66b3f98ba2bec4879b98e4fba0b3c0
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 9 13:41:31 2015 -0500
Revert "Rearrange if/else to make intent clearer"
This reverts commit 59b63c533fcaddcdfe4656e86a6f6c4cb1bc4a00.
commit 2874791b9c9cd807113e75e38be465f3785c154e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 9 13:38:13 2015 -0500
Revert "Move state into ProfileCollectorBuilder"
This reverts commit 0bb3ee0dd96170b06f07ec9e2435423d686a5ae6.
commit 0bb3ee0dd96170b06f07ec9e2435423d686a5ae6
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Dec 3 11:21:55 2015 -0500
Move state into ProfileCollectorBuilder
commit 59b63c533fcaddcdfe4656e86a6f6c4cb1bc4a00
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 2 17:21:12 2015 -0500
Rearrange if/else to make intent clearer
commit 511db0af2f3a86328028b88a6b25fa3dfbab963b
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 2 17:12:06 2015 -0500
Rename WEIGHT -> BUILD_WEIGHT
commit 02cc31767fb76a7ecd44a302435e93a05fb4220e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 2 17:11:22 2015 -0500
Move some of the collector wrapping logic into ProfileCollectorBuilder
commit e69356d3cb4c60fa281dad36d84faa64f5c32bc4
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 15:12:35 2015 -0500
Cleanup imports
commit c1b4f284f16712be60cd881f7e4a3e8175667d62
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 15:11:25 2015 -0500
Review cleanup: Make InternalProfileShardResults writeable
commit 9e61c72f7e1787540f511777050a572b7d297636
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 15:01:22 2015 -0500
Review cleanup: Merge ProfileShardResult, InternalProfileShardResult. Convert to writeable
commit 709184e1554f567c645690250131afe8568a5799
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 14:38:08 2015 -0500
Review cleanup: Merge ProfileResult, InternalProfileResult. Convert to writeable
commit 7d72690c44f626c34e9c608754bc7843dd7fd8fe
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 14:01:34 2015 -0500
Review cleanup: use primitive (and default) for profile flag
commit 97d557388541bbd3388cdcce7d9718914d88de6d
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 13:09:12 2015 -0500
Review cleanup: Use Collections.emptyMap() instead of building an empty one explicitly
commit 219585b8729a8b0982e653d99eb959efd0bef84e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 13:08:12 2015 -0500
Add todo to revisit profiler architecture in the future
commit b712edb2160e032ee4b2f2630fadf131a0936886
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 13:05:32 2015 -0500
Split collector serialization from timing, use writeable instead of streamable
Previously, the collector timing was done in the same class that was serialized, which required
leaving the collector null when serializing. Besides being a bit gross, this made it difficult to
change the class to Writeable.
This splits up the timing (InternalProfileCollector + ProfileCollector) and the serialization of
the times (CollectorResult). CollectorResult is writeable, and also acts as the public interface.
commit 6ddd77d066262d4400e3d338b11cebe7dd27ca78
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Nov 25 13:15:12 2015 -0500
Remove dead code
commit 06033f8a056e2121d157654a65895c82bbe93a51
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Nov 25 12:49:51 2015 -0500
Review cleanup: Delegate to in.getProfilers()
Note: Need to investigate how this is used exactly so we can add a test, it isn't touched by a
normal inner_hits query...
commit a77e13da21b4bad1176ca2b5d5b76034fb12802f
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Nov 25 11:59:58 2015 -0500
Review cleanup: collapse to single `if` statement
commit e97bb6215a5ebb508b0293ac3acd60d5ae479be1
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Nov 25 11:39:43 2015 -0500
Review cleanup: Return empty map instead of null for profile results
Note: we still need to check for nullness in SearchPhaseController, since an empty/no-hits result
won't have profiling instantiated (or any other component like aggs or suggest). Therefore
QuerySearchResult.profileResults() is still @Nullable
commit db8e691de2a727389378b459fa76c942572e6015
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Nov 25 10:14:47 2015 -0500
Review cleanup: renaming, formatting fixes, assertions
commit 9011775fe80ba22c2fd948ca64df634b4e32772d
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Nov 19 20:09:52 2015 -0500
[DOCS] Add missing annotation
commit 4b58560b06f08d4b99b149af20916ee839baabd7
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Nov 19 20:07:17 2015 -0500
[DOCS] Update documentation for new format
commit f0458c58e5538ed8ec94849d4baf3250aa9ec841
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Nov 17 10:14:09 2015 +0100
Reduce visibility of internal classes.
commit d0a7d319098e60b028fa772bf8a99b2df9cf6146
Merge: e158070 1bdf29e
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Nov 17 10:09:18 2015 +0100
Merge branch 'master' into query_profiler
commit e158070a48cb096551f3bb3ecdcf2b53bbc5e3c5
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Nov 17 10:08:48 2015 +0100
Fix compile error due to bad html in javadocs.
commit a566b5d08d659daccb087a9afbe908ec3d96cd6e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 16 17:48:37 2015 -0500
Remove unused collector
commit 4060cd72d150cc68573dbde62ca7321c47f75703
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 16 17:48:10 2015 -0500
Comment cleanup
commit 43137952bf74728f5f5d5a8d1bfc073e0f9fe4f9
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 16 17:32:06 2015 -0500
Fix negative formatted time
commit 5ef3a980266326aff12d4fe380f73455ff28209f
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 17:10:17 2015 +0100
Fix javadocs.
commit 276114d29e4b17a0cc0982cfff51434f712dc59e
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 16:25:23 2015 +0100
Fix: include rewrite time as well...
commit 21d9e17d05487bf4903ae3d2ab6f429bece2ffef
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 15:10:15 2015 +0100
Remove TODO about profiling explain.
commit 105a31e8e570efb879447159c3852871f5cf7db4
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 14:59:30 2015 +0100
Fix nocommit now that the global collector is a root collector.
commit 2e8fc5cf84adb1bfaba296808c329e5f982c9635
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 14:53:38 2015 +0100
Make collector wrapping more explicit/robust (and a bit less magical).
commit 5e30b570b0835e1ce79a57933a31b6a2d0d58e2d
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 12:44:03 2015 +0100
Simplify recording API a bit.
commit 9b453afced6adc0a59ca1d67d90c28796b105185
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 10:54:25 2015 +0100
Fix serialization-related nocommits.
commit ad97b200bb123d4e9255e7c8e02f7e43804057a5
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 10:46:30 2015 +0100
Fix DFS.
commit a6de06986cd348a831bd45e4f524d2e14d9e03c3
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Nov 12 19:29:16 2015 +0100
Remove forbidden @Test annotation.
commit 4991a28e19501109af98026e14756cb25a56f4f4
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Nov 12 19:25:59 2015 +0100
Limit the impact of query profiling on the SearchContext API.
Rule is: we can put as much as we want in the search.profile package but should
aim at touching as little as possible other areas of the code base.
commit 353d8d75a5ce04d9c3908a0a63d4ca6e884c519a
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Nov 12 18:05:09 2015 +0100
Remove dead code.
commit a3ffafb5ddbb5a2acf43403c946e5ed128f47528
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Nov 12 15:30:35 2015 +0100
Remove call to forbidden String.toLowerCase() API.
commit 1fa8c7a00324fa4e32bd24135ebba5ecf07606f1
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Nov 12 15:30:27 2015 +0100
Fix compilation.
commit 2067f1797e53bef0e1a8c9268956bc5fb8f8ad97
Merge: 22e631f fac472f
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Nov 12 15:21:12 2015 +0100
Merge branch 'master' into query_profiler
commit 22e631fe6471fed19236578e97c628d5cda401a9
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Nov 3 18:52:05 2015 -0500
Fix and simplify serialization of shard profile results
commit 461da250809451cd2b47daf647343afbb4b327f2
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Nov 3 18:32:22 2015 -0500
Remove helper methods, simpler without them
commit 5687aa1c93d45416d895c2eecc0e6a6b302139f2
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Nov 3 18:29:32 2015 -0500
[TESTS] Fix tests for new rewrite format
commit ba9e82857fc6d4c7b72ef4d962d2102459365299
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 30 15:28:14 2015 -0400
Rewrites begone! Record all rewrites as a single time metric
commit 5f28d7cdff9ee736651d564f71f713bf45fb1d91
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 29 15:36:06 2015 -0400
Merge duplicate rewrites into one entry
By using the Query as the key in a map, we can easily merge rewrites together. This means
the preProcess(), assertion and main query rewrites all get merged together. Downside is that
rewrites of the same Query (hashcode) but in different places get lumped together. I think the
simplicity of the solution is better than the slight loss in output fidelity.
commit 9a601ea46bb21052746157a45dcc6de6bc350e9c
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 29 15:28:27 2015 -0400
Allow multiple "searches" per profile (e.g. query + global agg)
commit ee30217328381cd83f9e653d3a4d870c1d2bdfce
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 29 11:29:18 2015 -0400
Update comment, add nullable annotation
commit 405c6463a64e118f170959827931e8c6a1661f13
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 29 11:04:30 2015 -0400
remove out-dated comment
commit 2819ae8f4cf1bfd5670dbd1c0e06195ae457b58f
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Oct 27 19:50:47 2015 +0100
Don't render children in the profiles when there are no children.
commit 7677c2ddefef321bbe74660471603d202a4ab66f
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Oct 27 19:50:35 2015 +0100
Set the profiler on the ContextIndexSearcher.
commit 74a4338c35dfed779adc025ec17cfd4d1c9f66f5
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Oct 27 19:50:01 2015 +0100
Fix json rendering.
commit 6674d5bebe187b0b0d8b424797606fdf2617dd27
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Oct 27 19:20:19 2015 +0100
Revert "nocommit - profiling never enabled because setProfile() on ContextIndexSearcher never called"
This reverts commit d3dc10949024342055f0d4fb7e16c7a43423bfab.
commit d3dc10949024342055f0d4fb7e16c7a43423bfab
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 17:20:57 2015 -0400
nocommit - profiling never enabled because setProfile() on ContextIndexSearcher never called
Previously, it was enabled by using DefaultSearchContext as a third-party "proxy", but since
the refactor to make it unit testable, setProfile() needs to be called explicitly. Unfortunately,
this is not possible because SearchService only has access to an IndexSearcher. And it is not
cast'able to a DefaultIndexSearcher.
commit b9ba9c5d1f93b9c45e97b0a4e35da6f472c9ea53
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 16:27:00 2015 -0400
[TESTS] Fix unit tests
commit cf5d1e016b2b4a583175e07c16c7152f167695ce
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 16:22:34 2015 -0400
Increment token after recording a rewrite
commit b7d08f64034e498533c4a81bff8727dd8ac2843e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 16:14:09 2015 -0400
Fix NPE if a top-level root doesn't have children
commit e4d3b514bafe2a3a9db08438c89f0ed68628f2d6
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 16:05:47 2015 -0400
Fix NPE when profiling is disabled
commit 445384fe48ed62fdd01f7fc9bf3e8361796d9593
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 16:05:37 2015 -0400
[TESTS] Fix integration tests
commit b478296bb04fece827a169e7522df0a5ea7840a3
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 15:43:24 2015 -0400
Move rewrites to their own section, remove reconciliation
Big commit because the structural change affected a lot of the wrapping code. Major changes:
- Rewrites are now in their own section in the response
- Reconciliation is gone...we don't attempt to roll the rewrites into the query tree structure
- InternalProfileShardResults (plural) simply holds a Map<String, InternalProfileShardResult> and
helps to serialize / ToXContent
- InternalProfileShardResult (singular) is now the holder for all shard-level profiling details. Currently
this includes query, collectors and rewrite. In the future it would hold suggest, aggs, etc
- When the user requests the profiled results, they get back a Map<String, ProfileShardResult>
instead of doing silly helper methods to convert to maps, etc
- Shard details are baked into a string instead of serializing the object
commit 24819ad094b208d0e94f17ce9c3f7c92f7414124
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 10:25:38 2015 -0400
Make Profile results immutable by removing relative_time
commit bfaf095f45fed74194ef78160a8e5dcae1850f9e
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 23 10:54:59 2015 +0200
Add nocommits.
commit e9a128d0d26d5b383b52135ca886f2c987850179
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 23 10:39:37 2015 +0200
Move all profile-related classes to the same package.
commit f20b7c7fdf85384ecc37701bb65310fb8c20844f
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 23 10:33:14 2015 +0200
Reorganize code a bit to ease unit testing of ProfileCollector.
commit 3261306edad6a0c70f59eaee8fe58560f61a75fd
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 18:07:28 2015 +0200
Remove irrelevant nocommit.
commit a6ac868dad12a2e17929878681f66dbd0948d322
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 18:06:45 2015 +0200
Make CollectorResult's reason a free-text field to ease bw compat.
commit 5d0bf170781a950d08b81871cd1e403e49f3cc12
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 16:50:52 2015 +0200
Add unit tests for ProfileWeight/ProfileScorer.
commit 2cd88c412c6e62252504ef69a59216adbb574ce4
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 15:55:17 2015 +0200
Rename InternalQueryProfiler to Profiler.
commit 84f5718fa6779f710da129d9e0e6ff914fd85e36
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 15:53:58 2015 +0200
Merge InternalProfileBreakdown into ProfileBreakdown.
commit 135168eaeb8999c8117ea25288104b0961ce9b35
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 13:56:57 2015 +0200
Make it possible to instantiate a ContextIndexSearcher without SearchContext.
commit 5493fb52376b48460c4ce2dedbe00cc5f6620499
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 11:53:29 2015 +0200
Move ProfileWeight/Scorer to their own files.
commit bf2d917b9dae3b32dfc29c35a7cac4ccb7556cce
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 11:38:24 2015 +0200
Fix bug that caused phrase queries to fail.
commit b2bb0c92c343334ec1703a221af24a1b55e36d53
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 11:36:17 2015 +0200
Parsing happens on the coordinating node now.
commit 416cabb8621acb5cd8dfa77374fd23e428f52fe9
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 11:22:17 2015 +0200
Fix compilation (in particular remove guava deps).
commit f996508645f842629d403fc2e71c1890c0e2cac9
Merge: 4616a25 bc3b91e
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 10:44:38 2015 +0200
Merge branch 'master' into query_profiler
commit 4616a25afffe9c24c6531028f7fccca4303d2893
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Oct 20 12:11:32 2015 -0400
Make Java Count API compatible with profiling
commit cbfba74e16083d719722500ac226efdb5cb2ff55
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Oct 20 12:11:19 2015 -0400
Fix serialization of profile query param, NPE
commit e33ffac383b03247046913da78c8a27e457fae78
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Oct 20 11:17:48 2015 -0400
TestSearchContext should return null Profiler instead of exception
commit 73a02d69b466dc1a5b8a5f022464d6c99e6c2ac3
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Oct 19 12:07:29 2015 -0400
[DOCS] Update docs to reflect new ID format
commit 36248e388c354f954349ecd498db7b66f84ce813
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Oct 19 12:03:03 2015 -0400
Use the full [node][index][shard] string as profile result ID
commit 5cfcc4a6a6b0bcd6ebaa7c8a2d0acc32529a80e1
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 15 17:51:40 2015 -0400
Add failing test for phrase matching
Stack trace generated:
[2015-10-15 17:50:54,438][ERROR][org.elasticsearch.search.profile] shard [[JNj7RX_oSJikcnX72aGBoA][test][2]], reason [RemoteTransportException[[node_s0][local[1]][indices:data/read/search[phase/query]]]; nested: QueryPhaseExecutionException[Query Failed [Failed to execute main query]]; nested: AssertionError[nextPosition() called more than freq() times!]; ], cause [java.lang.AssertionError: nextPosition() called more than freq() times!
at org.apache.lucene.index.AssertingLeafReader$AssertingPostingsEnum.nextPosition(AssertingLeafReader.java:353)
at org.apache.lucene.search.ExactPhraseScorer.phraseFreq(ExactPhraseScorer.java:132)
at org.apache.lucene.search.ExactPhraseScorer.access$000(ExactPhraseScorer.java:27)
at org.apache.lucene.search.ExactPhraseScorer$1.matches(ExactPhraseScorer.java:69)
at org.elasticsearch.common.lucene.search.ProfileQuery$ProfileScorer$2.matches(ProfileQuery.java:226)
at org.apache.lucene.search.ConjunctionDISI$TwoPhaseConjunctionDISI.matches(ConjunctionDISI.java:175)
at org.apache.lucene.search.ConjunctionDISI$TwoPhase.matches(ConjunctionDISI.java:213)
at org.apache.lucene.search.ConjunctionDISI.doNext(ConjunctionDISI.java:128)
at org.apache.lucene.search.ConjunctionDISI.nextDoc(ConjunctionDISI.java:151)
at org.apache.lucene.search.ConjunctionScorer.nextDoc(ConjunctionScorer.java:62)
at org.elasticsearch.common.lucene.search.ProfileQuery$ProfileScorer$1.nextDoc(ProfileQuery.java:205)
at org.apache.lucene.search.Weight$DefaultBulkScorer.scoreAll(Weight.java:224)
at org.apache.lucene.search.Weight$DefaultBulkScorer.score(Weight.java:169)
at org.apache.lucene.search.BulkScorer.score(BulkScorer.java:39)
at org.apache.lucene.search.IndexSearcher.search(IndexSearcher.java:795)
at org.apache.lucene.search.IndexSearcher.search(IndexSearcher.java:509)
at org.elasticsearch.search.query.QueryPhase.execute(QueryPhase.java:347)
at org.elasticsearch.search.query.QueryPhase.execute(QueryPhase.java:111)
at org.elasticsearch.search.SearchService.loadOrExecuteQueryPhase(SearchService.java:366)
at org.elasticsearch.search.SearchService.executeQueryPhase(SearchService.java:378)
at org.elasticsearch.search.action.SearchServiceTransportAction$SearchQueryTransportHandler.messageReceived(SearchServiceTransportAction.java:368)
at org.elasticsearch.search.action.SearchServiceTransportAction$SearchQueryTransportHandler.messageReceived(SearchServiceTransportAction.java:365)
at org.elasticsearch.transport.local.LocalTransport$2.doRun(LocalTransport.java:280)
at org.elasticsearch.common.util.concurrent.AbstractRunnable.run(AbstractRunnable.java:37)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
at java.lang.Thread.run(Thread.java:745)
commit 889fe6383370fe919aaa9f0af398e3040209e40b
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 15 17:30:38 2015 -0400
[DOCS] More docs
commit 89177965d031d84937753538b88ea5ebae2956b0
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 15 09:59:09 2015 -0400
Fix multi-stage rewrites to recursively find most appropriate descendant rewrite
Previously, we chose the first rewrite that matched. But in situations where a query may
rewrite several times, this won't build the tree correctly. Instead we need to recurse
down all the rewrites until we find the most appropriate "leaf" rewrite
The implementation of this is kinda gross: we recursively call getRewrittenParentToken(),
which does a linear scan over the rewriteMap and tries to find rewrites with a larger token
value (since we know child tokens are always larger). Can almost certainly find a better
way to do this...
commit 0b4d782b5348e5d03fd26f7d91bc4a3fbcb7f6a5
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Oct 14 19:30:06 2015 -0400
[Docs] Documentation checkpoint
commit 383636453f6610fcfef9070c21ae7ca11346793e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Sep 16 16:02:22 2015 -0400
Comments
commit a81e8f31e681be16e89ceab9ba3c3e0a018f18ef
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Sep 16 15:48:49 2015 -0400
[TESTS] Ensure all tests use QUERY_THEN_FETCH, DFS does not profile
commit 1255c2d790d85fcb9cbb78bf2a53195138c6bc24
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Sep 15 16:43:46 2015 -0400
Refactor rewrite handling to handle identical rewrites
commit 85b7ec82eb0b26a6fe87266b38f5f86f9ac0c44f
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Sep 15 08:51:14 2015 -0400
Don't update parent when a token is added as root -- Fixes NPE
commit 109d02bdbc49741a3b61e8624521669b0968b839
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Sep 15 08:50:40 2015 -0400
Don't set the rewritten query if not profiling -- Fixes NPE
commit 233cf5e85f6f2c39ed0a2a33d7edd3bbd40856e8
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Sep 14 18:04:51 2015 -0400
Update tests to new response format
commit a930b1fc19de3a329abc8ffddc6711c1246a4b15
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Sep 14 18:03:58 2015 -0400
Fix serialization
commit 69afdd303660510c597df9bada5531b19d134f3d
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Sep 14 15:11:31 2015 -0400
Comments and cleanup
commit 64e7ca7f78187875378382ec5d5aa2462ff71df5
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Sep 14 14:40:21 2015 -0400
Move timing into dedicated class, add proper rewrite integration
commit b44ff85ddbba0a080e65f2e7cc8c50d30e95df8e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Sep 14 12:00:38 2015 -0400
Checkpoint - Refactoring to use a token-based dependency tree
commit 52cedd5266d6a87445c6a4cff3be8ff2087cd1b7
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Sep 4 19:18:19 2015 -0400
Need to set context profiling flag before calling queryPhase.preProcess
commit c524670cb1ce29b4b3a531fa2bff0c403b756f46
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Sep 4 18:00:37 2015 +0200
Reduce profiling overhead a bit.
This removes hash-table lookups everytime we start/stop a profiling clock.
commit 111444ff8418737082236492b37321fc96041e09
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Sep 4 16:18:59 2015 +0200
Add profiling of two-phase iterators.
This is useful for eg. phrase queries or script filters, since they are
typically consumed through their two-phase iterator instead of the scorer.
commit f275e690459e73211bc8494c6de595c0320f4c0b
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Sep 4 16:03:21 2015 +0200
Some more improvements.
I changed profiling to disable bulk scoring, since it makes it impossible to
know where time is spent. Also I removed profiling of operations that are
always fast (eg. normalization) and added nextDoc/advance.
commit 3c8dcd872744de8fd76ce13b6f18f36f8de44068
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Sep 4 14:39:50 2015 +0200
Remove println.
commit d68304862fb38a3823aebed35a263bd9e2176c2f
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Sep 4 14:36:03 2015 +0200
Fix some test failures introduced by the rebase...
commit 04d53ca89fb34b7a21515d770c32aaffcc513b90
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Sep 4 13:57:35 2015 +0200
Reconcile conflicting changes after rebase
commit fed03ec8e2989a0678685cd6c50a566cec42ea4f
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Aug 20 22:40:39 2015 -0400
Add Collectors to profile results
Profile response element has now been re-arranged so that everything is listed per-shard to
facilitate grouping elements together. The new `collector` element looks like this:
```
"profile": {
"shards": [
{
"shard_id": "keP4YFywSXWALCl4m4k24Q",
"query": [...],
"collector": [
{
"name": "MultiCollector",
"purpose": "search_multi",
"time": "16.44504400ms",
"relative_time": "100.0000000%",
"children": [
{
"name": "FilteredCollector",
"purpose": "search_post_filter",
"time": "4.556013000ms",
"relative_time": "27.70447437%",
"children": [
{
"name": "SimpleTopScoreDocCollector",
"purpose": "search_sorted",
"time": "1.352166000ms",
"relative_time": "8.222331299%",
"children": []
}
]
},
{
"name": "BucketCollector: [[non_global_term, another_agg]]",
"purpose": "aggregation",
"time": "10.40379400ms",
"relative_time": "63.26400829%",
"children": []
},
...
```
commit 1368b495c934be642c00f6cbf9fc875d7e6c07ff
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Aug 19 12:43:03 2015 -0400
Move InternalProfiler to profile package
commit 53584de910db6d4a6bb374c9ebb954f204882996
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 18:34:58 2015 -0400
Only reconcile rewrite timing when rewritten query is different from original
commit 9804c3b29d2107cd97f1c7e34d77171b62cb33d0
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 16:40:15 2015 -0400
Comments and cleanup
commit 8e898cc7c59c0c1cc5ed576dfed8e3034ca0967f
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 14:19:07 2015 -0400
[TESTS] Fix comparison test to ensure results sort identically
commit f402a29001933eef29d5a62e81c8563f1c8d0969
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 14:17:59 2015 -0400
Add note about DFS being unable to profile
commit d446e08d3bc91cd85b24fc908e2d82fc5739d598
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 14:17:23 2015 -0400
Implement some missing methods
commit 13ca94fb86fb037a30d181b73d9296153a63d6e4
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 13:10:54 2015 -0400
[TESTS] Comments & cleanup
commit c76c8c771fdeee807761c25938a642612a6ed8e7
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 13:06:08 2015 -0400
[TESTS] Fix profileMatchesRegular to handle NaN scores and nearlyEqual floats
commit 7e7a10ecd26677b2239149468e24938ce5cc18e1
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 12:22:16 2015 -0400
Move nearlyEquals() utility function to shared location
commit 842222900095df4b27ff3593dbb55a42549f2697
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 12:04:35 2015 -0400
Fixup rebase conflicts
commit 674f162d7704dd2034b8361358decdefce1f76ce
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Aug 17 15:29:35 2015 -0400
[TESTS] Update match and bool tests
commit 520380a85456d7137734aed0b06a740e18c9cdec
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Aug 17 15:28:09 2015 -0400
Make naming consistent re: plural
commit b9221501d839bb24d6db575d08e9bee34043fc65
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Aug 17 15:27:39 2015 -0400
Children need to be added to list after serialization
commit 05fa51df940c332fbc140517ee56e849f2d40a72
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Aug 17 15:22:41 2015 -0400
Re-enable bypass for non-profiled queries
commit f132204d264af77a75bd26a02d4e251a19eb411d
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Aug 17 15:21:14 2015 -0400
Fix serialization of QuerySearchResult, InternalProfileResult
commit 27b98fd475fc2e9508c91436ef30624bdbee54ba
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Aug 10 17:39:17 2015 -0400
Start to add back tests, refactor Java api
commit bcfc9fefd49307045108408dc160774666510e85
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 4 17:08:10 2015 -0400
Checkpoint
commit 26a530e0101ce252450eb23e746e48c2fd1bfcae
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Jul 14 13:30:32 2015 -0400
Add createWeight() checkpoint
commit f0dd61de809c5c13682aa213c0be65972537a0df
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Jul 13 12:36:27 2015 -0400
checkpoint
commit 377ee8ce5729b8d388c4719913b48fae77a16686
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Mar 18 10:45:01 2015 -0400
checkpoint
2015-12-17 12:05:43 -05:00
|
|
|
import org.elasticsearch.search.profile.Profilers;
|
2015-12-18 12:43:47 -08:00
|
|
|
import org.elasticsearch.search.query.QueryPhaseExecutionException;
|
2013-08-26 16:28:27 +02:00
|
|
|
import org.elasticsearch.search.query.QuerySearchResult;
|
2017-08-25 13:46:57 -04:00
|
|
|
import org.elasticsearch.search.rescore.RescoreContext;
|
2016-06-08 15:28:06 +02:00
|
|
|
import org.elasticsearch.search.slice.SliceBuilder;
|
2016-04-25 17:45:33 +02:00
|
|
|
import org.elasticsearch.search.sort.SortAndFormats;
|
2013-08-26 16:28:27 +02:00
|
|
|
import org.elasticsearch.search.suggest.SuggestionSearchContext;
|
|
|
|
|
2015-08-14 13:29:33 +02:00
|
|
|
import java.io.IOException;
|
2016-10-14 16:26:09 +02:00
|
|
|
import java.io.UncheckedIOException;
|
2015-12-18 12:43:47 -08:00
|
|
|
import java.util.ArrayList;
|
2017-05-04 16:29:35 +02:00
|
|
|
import java.util.Arrays;
|
2015-12-18 12:43:47 -08:00
|
|
|
import java.util.Collections;
|
|
|
|
import java.util.HashMap;
|
|
|
|
import java.util.List;
|
|
|
|
import java.util.Map;
|
2019-02-21 12:42:54 +02:00
|
|
|
import java.util.function.LongSupplier;
|
2013-08-26 16:28:27 +02:00
|
|
|
|
2016-09-12 22:42:55 +02:00
|
|
|
final class DefaultSearchContext extends SearchContext {
|
2013-08-26 16:28:27 +02:00
|
|
|
|
2020-03-09 11:59:30 -04:00
|
|
|
private final SearchContextId id;
|
2013-08-26 16:28:27 +02:00
|
|
|
private final ShardSearchRequest request;
|
|
|
|
private final SearchShardTarget shardTarget;
|
2019-02-21 12:42:54 +02:00
|
|
|
private final LongSupplier relativeTimeSupplier;
|
2013-08-26 16:28:27 +02:00
|
|
|
private SearchType searchType;
|
|
|
|
private final Engine.Searcher engineSearcher;
|
2014-02-26 16:40:15 +01:00
|
|
|
private final BigArrays bigArrays;
|
2013-08-26 16:28:27 +02:00
|
|
|
private final IndexShard indexShard;
|
2018-04-26 09:58:17 +02:00
|
|
|
private final ClusterService clusterService;
|
2013-08-26 16:28:27 +02:00
|
|
|
private final IndexService indexService;
|
|
|
|
private final ContextIndexSearcher searcher;
|
|
|
|
private final DfsSearchResult dfsResult;
|
|
|
|
private final QuerySearchResult queryResult;
|
|
|
|
private final FetchSearchResult fetchResult;
|
2016-11-07 17:20:06 +09:00
|
|
|
private final float queryBoost;
|
2020-03-23 21:30:34 +01:00
|
|
|
private final boolean lowLevelCancellation;
|
2016-06-27 18:41:18 -04:00
|
|
|
private TimeValue timeout;
|
2014-07-18 10:53:20 -04:00
|
|
|
// terminate after count
|
|
|
|
private int terminateAfter = DEFAULT_TERMINATE_AFTER;
|
2013-08-26 16:28:27 +02:00
|
|
|
private List<String> groupStats;
|
2015-08-18 17:01:47 +02:00
|
|
|
private ScrollContext scrollContext;
|
2013-08-26 16:28:27 +02:00
|
|
|
private boolean explain;
|
|
|
|
private boolean version = false; // by default, we don't return versions
|
2019-01-23 09:01:58 +01:00
|
|
|
private boolean seqAndPrimaryTerm = false;
|
2016-08-17 15:59:38 +02:00
|
|
|
private StoredFieldsContext storedFields;
|
2013-08-26 16:28:27 +02:00
|
|
|
private ScriptFieldsContext scriptFields;
|
|
|
|
private FetchSourceContext fetchSourceContext;
|
2020-02-13 12:59:59 -08:00
|
|
|
private FetchDocValuesContext docValuesContext;
|
2020-07-28 10:58:20 -07:00
|
|
|
private FetchFieldsContext fetchFieldsContext;
|
2013-08-26 16:28:27 +02:00
|
|
|
private int from = -1;
|
|
|
|
private int size = -1;
|
2016-04-25 17:45:33 +02:00
|
|
|
private SortAndFormats sort;
|
2013-08-26 16:28:27 +02:00
|
|
|
private Float minimumScore;
|
|
|
|
private boolean trackScores = false; // when sorting, track scores as well...
|
2019-01-04 20:36:49 +01:00
|
|
|
private int trackTotalHitsUpTo = SearchContext.DEFAULT_TRACK_TOTAL_HITS_UP_TO;
|
2016-01-12 17:40:34 +01:00
|
|
|
private FieldDoc searchAfter;
|
2017-01-23 16:33:51 +01:00
|
|
|
private CollapseContext collapse;
|
2016-05-10 12:10:55 +02:00
|
|
|
// filter for sliced scroll
|
2016-06-08 15:28:06 +02:00
|
|
|
private SliceBuilder sliceBuilder;
|
2019-11-28 18:23:09 +01:00
|
|
|
private SearchShardTask task;
|
2018-04-26 09:58:17 +02:00
|
|
|
private final Version minNodeVersion;
|
2016-10-25 12:27:33 -10:00
|
|
|
|
2016-01-06 12:12:20 -05:00
|
|
|
/**
|
|
|
|
* The original query as sent by the user without the types and aliases
|
|
|
|
* applied. Putting things in here leaks them into highlighting so don't add
|
|
|
|
* things like the type filter or alias filters.
|
|
|
|
*/
|
2013-08-26 16:28:27 +02:00
|
|
|
private ParsedQuery originalQuery;
|
2016-09-12 22:42:55 +02:00
|
|
|
|
2016-01-06 12:12:20 -05:00
|
|
|
/**
|
|
|
|
* The query to actually execute.
|
|
|
|
*/
|
2013-08-26 16:28:27 +02:00
|
|
|
private Query query;
|
2015-05-05 08:27:52 +02:00
|
|
|
private ParsedQuery postFilter;
|
|
|
|
private Query aliasFilter;
|
2013-08-26 16:28:27 +02:00
|
|
|
private int[] docIdsToLoad;
|
|
|
|
private int docsIdsToLoadFrom;
|
|
|
|
private int docsIdsToLoadSize;
|
2013-11-24 03:13:08 -08:00
|
|
|
private SearchContextAggregations aggregations;
|
2020-07-29 11:39:17 -07:00
|
|
|
private SearchHighlightContext highlight;
|
2013-08-26 16:28:27 +02:00
|
|
|
private SuggestionSearchContext suggest;
|
2017-08-25 13:46:57 -04:00
|
|
|
private List<RescoreContext> rescore;
|
2013-08-26 16:28:27 +02:00
|
|
|
private volatile long keepAlive;
|
2015-07-20 12:42:13 -04:00
|
|
|
private final long originNanoTime = System.nanoTime();
|
2014-02-18 14:43:25 -05:00
|
|
|
private volatile long lastAccessTime = -1;
|
Add ability to profile query and collectors
Provides a new flag which can be enabled on a per-request basis.
When `"profile": true` is set, the search request will execute in a
mode that collects detailed timing information for query components.
```
GET /test/test/_search
{
"profile": true,
"query": {
"match": {
"foo": "bar"
}
}
}
```
Closes #14889
Squashed commit of the following:
commit a92db5723d2c61b8449bd163d2f006d12f9889ad
Merge: 117dd99 3f87b08
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Dec 17 09:44:10 2015 -0500
Merge remote-tracking branch 'upstream/master' into query_profiler
commit 117dd9992e8014b70203c6110925643769d80e62
Merge: 9b29d68 82a64fd
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Dec 15 13:27:18 2015 -0500
Merge remote-tracking branch 'upstream/master' into query_profiler
Conflicts:
core/src/main/java/org/elasticsearch/search/SearchService.java
commit 9b29d6823a71140ecd872df25ff9f7478e7fe766
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Dec 14 16:13:23 2015 -0500
[TEST] Profile flag needs to be set, ensure searches go against primary only for consistency
commit 4d602d8ad1f8cbc7b475450921fa3bc7d395b34f
Merge: 8b48e87 7742c1e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Dec 14 10:56:25 2015 -0500
Merge remote-tracking branch 'upstream/master' into query_profiler
commit 8b48e876348b163ab730eeca7fa35142165b05f9
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Dec 14 10:56:01 2015 -0500
Delegate straight to in.matchCost, no need for profiling
commit fde3b0587911f0b5f15e779c671d0510cbd568a9
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Dec 14 10:28:23 2015 -0500
Documentation tweaks, renaming build_weight -> create_weight
commit 46f5e011ee23fe9bb8a1f11ceb4fa9d19fe48e2e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Dec 14 10:27:52 2015 -0500
Profile TwoPhaseIterator should override matchCost()
commit b59f894ddb11b2a7beebba06c4ec5583ff91a7b2
Merge: 9aa1a3a b4e0c87
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 9 14:23:26 2015 -0500
Merge remote-tracking branch 'upstream/master' into query_profiler
commit 9aa1a3a25c34c9cd9fffaa6114c25a0ec791307d
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 9 13:41:48 2015 -0500
Revert "Move some of the collector wrapping logic into ProfileCollectorBuilder"
This reverts commit 02cc31767fb76a7ecd44a302435e93a05fb4220e.
commit 57f7c04cea66b3f98ba2bec4879b98e4fba0b3c0
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 9 13:41:31 2015 -0500
Revert "Rearrange if/else to make intent clearer"
This reverts commit 59b63c533fcaddcdfe4656e86a6f6c4cb1bc4a00.
commit 2874791b9c9cd807113e75e38be465f3785c154e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 9 13:38:13 2015 -0500
Revert "Move state into ProfileCollectorBuilder"
This reverts commit 0bb3ee0dd96170b06f07ec9e2435423d686a5ae6.
commit 0bb3ee0dd96170b06f07ec9e2435423d686a5ae6
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Dec 3 11:21:55 2015 -0500
Move state into ProfileCollectorBuilder
commit 59b63c533fcaddcdfe4656e86a6f6c4cb1bc4a00
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 2 17:21:12 2015 -0500
Rearrange if/else to make intent clearer
commit 511db0af2f3a86328028b88a6b25fa3dfbab963b
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 2 17:12:06 2015 -0500
Rename WEIGHT -> BUILD_WEIGHT
commit 02cc31767fb76a7ecd44a302435e93a05fb4220e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 2 17:11:22 2015 -0500
Move some of the collector wrapping logic into ProfileCollectorBuilder
commit e69356d3cb4c60fa281dad36d84faa64f5c32bc4
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 15:12:35 2015 -0500
Cleanup imports
commit c1b4f284f16712be60cd881f7e4a3e8175667d62
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 15:11:25 2015 -0500
Review cleanup: Make InternalProfileShardResults writeable
commit 9e61c72f7e1787540f511777050a572b7d297636
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 15:01:22 2015 -0500
Review cleanup: Merge ProfileShardResult, InternalProfileShardResult. Convert to writeable
commit 709184e1554f567c645690250131afe8568a5799
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 14:38:08 2015 -0500
Review cleanup: Merge ProfileResult, InternalProfileResult. Convert to writeable
commit 7d72690c44f626c34e9c608754bc7843dd7fd8fe
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 14:01:34 2015 -0500
Review cleanup: use primitive (and default) for profile flag
commit 97d557388541bbd3388cdcce7d9718914d88de6d
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 13:09:12 2015 -0500
Review cleanup: Use Collections.emptyMap() instead of building an empty one explicitly
commit 219585b8729a8b0982e653d99eb959efd0bef84e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 13:08:12 2015 -0500
Add todo to revisit profiler architecture in the future
commit b712edb2160e032ee4b2f2630fadf131a0936886
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 13:05:32 2015 -0500
Split collector serialization from timing, use writeable instead of streamable
Previously, the collector timing was done in the same class that was serialized, which required
leaving the collector null when serializing. Besides being a bit gross, this made it difficult to
change the class to Writeable.
This splits up the timing (InternalProfileCollector + ProfileCollector) and the serialization of
the times (CollectorResult). CollectorResult is writeable, and also acts as the public interface.
commit 6ddd77d066262d4400e3d338b11cebe7dd27ca78
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Nov 25 13:15:12 2015 -0500
Remove dead code
commit 06033f8a056e2121d157654a65895c82bbe93a51
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Nov 25 12:49:51 2015 -0500
Review cleanup: Delegate to in.getProfilers()
Note: Need to investigate how this is used exactly so we can add a test, it isn't touched by a
normal inner_hits query...
commit a77e13da21b4bad1176ca2b5d5b76034fb12802f
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Nov 25 11:59:58 2015 -0500
Review cleanup: collapse to single `if` statement
commit e97bb6215a5ebb508b0293ac3acd60d5ae479be1
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Nov 25 11:39:43 2015 -0500
Review cleanup: Return empty map instead of null for profile results
Note: we still need to check for nullness in SearchPhaseController, since an empty/no-hits result
won't have profiling instantiated (or any other component like aggs or suggest). Therefore
QuerySearchResult.profileResults() is still @Nullable
commit db8e691de2a727389378b459fa76c942572e6015
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Nov 25 10:14:47 2015 -0500
Review cleanup: renaming, formatting fixes, assertions
commit 9011775fe80ba22c2fd948ca64df634b4e32772d
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Nov 19 20:09:52 2015 -0500
[DOCS] Add missing annotation
commit 4b58560b06f08d4b99b149af20916ee839baabd7
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Nov 19 20:07:17 2015 -0500
[DOCS] Update documentation for new format
commit f0458c58e5538ed8ec94849d4baf3250aa9ec841
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Nov 17 10:14:09 2015 +0100
Reduce visibility of internal classes.
commit d0a7d319098e60b028fa772bf8a99b2df9cf6146
Merge: e158070 1bdf29e
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Nov 17 10:09:18 2015 +0100
Merge branch 'master' into query_profiler
commit e158070a48cb096551f3bb3ecdcf2b53bbc5e3c5
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Nov 17 10:08:48 2015 +0100
Fix compile error due to bad html in javadocs.
commit a566b5d08d659daccb087a9afbe908ec3d96cd6e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 16 17:48:37 2015 -0500
Remove unused collector
commit 4060cd72d150cc68573dbde62ca7321c47f75703
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 16 17:48:10 2015 -0500
Comment cleanup
commit 43137952bf74728f5f5d5a8d1bfc073e0f9fe4f9
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 16 17:32:06 2015 -0500
Fix negative formatted time
commit 5ef3a980266326aff12d4fe380f73455ff28209f
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 17:10:17 2015 +0100
Fix javadocs.
commit 276114d29e4b17a0cc0982cfff51434f712dc59e
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 16:25:23 2015 +0100
Fix: include rewrite time as well...
commit 21d9e17d05487bf4903ae3d2ab6f429bece2ffef
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 15:10:15 2015 +0100
Remove TODO about profiling explain.
commit 105a31e8e570efb879447159c3852871f5cf7db4
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 14:59:30 2015 +0100
Fix nocommit now that the global collector is a root collector.
commit 2e8fc5cf84adb1bfaba296808c329e5f982c9635
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 14:53:38 2015 +0100
Make collector wrapping more explicit/robust (and a bit less magical).
commit 5e30b570b0835e1ce79a57933a31b6a2d0d58e2d
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 12:44:03 2015 +0100
Simplify recording API a bit.
commit 9b453afced6adc0a59ca1d67d90c28796b105185
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 10:54:25 2015 +0100
Fix serialization-related nocommits.
commit ad97b200bb123d4e9255e7c8e02f7e43804057a5
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 10:46:30 2015 +0100
Fix DFS.
commit a6de06986cd348a831bd45e4f524d2e14d9e03c3
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Nov 12 19:29:16 2015 +0100
Remove forbidden @Test annotation.
commit 4991a28e19501109af98026e14756cb25a56f4f4
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Nov 12 19:25:59 2015 +0100
Limit the impact of query profiling on the SearchContext API.
Rule is: we can put as much as we want in the search.profile package but should
aim at touching as little as possible other areas of the code base.
commit 353d8d75a5ce04d9c3908a0a63d4ca6e884c519a
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Nov 12 18:05:09 2015 +0100
Remove dead code.
commit a3ffafb5ddbb5a2acf43403c946e5ed128f47528
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Nov 12 15:30:35 2015 +0100
Remove call to forbidden String.toLowerCase() API.
commit 1fa8c7a00324fa4e32bd24135ebba5ecf07606f1
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Nov 12 15:30:27 2015 +0100
Fix compilation.
commit 2067f1797e53bef0e1a8c9268956bc5fb8f8ad97
Merge: 22e631f fac472f
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Nov 12 15:21:12 2015 +0100
Merge branch 'master' into query_profiler
commit 22e631fe6471fed19236578e97c628d5cda401a9
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Nov 3 18:52:05 2015 -0500
Fix and simplify serialization of shard profile results
commit 461da250809451cd2b47daf647343afbb4b327f2
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Nov 3 18:32:22 2015 -0500
Remove helper methods, simpler without them
commit 5687aa1c93d45416d895c2eecc0e6a6b302139f2
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Nov 3 18:29:32 2015 -0500
[TESTS] Fix tests for new rewrite format
commit ba9e82857fc6d4c7b72ef4d962d2102459365299
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 30 15:28:14 2015 -0400
Rewrites begone! Record all rewrites as a single time metric
commit 5f28d7cdff9ee736651d564f71f713bf45fb1d91
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 29 15:36:06 2015 -0400
Merge duplicate rewrites into one entry
By using the Query as the key in a map, we can easily merge rewrites together. This means
the preProcess(), assertion and main query rewrites all get merged together. Downside is that
rewrites of the same Query (hashcode) but in different places get lumped together. I think the
simplicity of the solution is better than the slight loss in output fidelity.
commit 9a601ea46bb21052746157a45dcc6de6bc350e9c
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 29 15:28:27 2015 -0400
Allow multiple "searches" per profile (e.g. query + global agg)
commit ee30217328381cd83f9e653d3a4d870c1d2bdfce
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 29 11:29:18 2015 -0400
Update comment, add nullable annotation
commit 405c6463a64e118f170959827931e8c6a1661f13
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 29 11:04:30 2015 -0400
remove out-dated comment
commit 2819ae8f4cf1bfd5670dbd1c0e06195ae457b58f
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Oct 27 19:50:47 2015 +0100
Don't render children in the profiles when there are no children.
commit 7677c2ddefef321bbe74660471603d202a4ab66f
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Oct 27 19:50:35 2015 +0100
Set the profiler on the ContextIndexSearcher.
commit 74a4338c35dfed779adc025ec17cfd4d1c9f66f5
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Oct 27 19:50:01 2015 +0100
Fix json rendering.
commit 6674d5bebe187b0b0d8b424797606fdf2617dd27
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Oct 27 19:20:19 2015 +0100
Revert "nocommit - profiling never enabled because setProfile() on ContextIndexSearcher never called"
This reverts commit d3dc10949024342055f0d4fb7e16c7a43423bfab.
commit d3dc10949024342055f0d4fb7e16c7a43423bfab
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 17:20:57 2015 -0400
nocommit - profiling never enabled because setProfile() on ContextIndexSearcher never called
Previously, it was enabled by using DefaultSearchContext as a third-party "proxy", but since
the refactor to make it unit testable, setProfile() needs to be called explicitly. Unfortunately,
this is not possible because SearchService only has access to an IndexSearcher. And it is not
cast'able to a DefaultIndexSearcher.
commit b9ba9c5d1f93b9c45e97b0a4e35da6f472c9ea53
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 16:27:00 2015 -0400
[TESTS] Fix unit tests
commit cf5d1e016b2b4a583175e07c16c7152f167695ce
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 16:22:34 2015 -0400
Increment token after recording a rewrite
commit b7d08f64034e498533c4a81bff8727dd8ac2843e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 16:14:09 2015 -0400
Fix NPE if a top-level root doesn't have children
commit e4d3b514bafe2a3a9db08438c89f0ed68628f2d6
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 16:05:47 2015 -0400
Fix NPE when profiling is disabled
commit 445384fe48ed62fdd01f7fc9bf3e8361796d9593
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 16:05:37 2015 -0400
[TESTS] Fix integration tests
commit b478296bb04fece827a169e7522df0a5ea7840a3
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 15:43:24 2015 -0400
Move rewrites to their own section, remove reconciliation
Big commit because the structural change affected a lot of the wrapping code. Major changes:
- Rewrites are now in their own section in the response
- Reconciliation is gone...we don't attempt to roll the rewrites into the query tree structure
- InternalProfileShardResults (plural) simply holds a Map<String, InternalProfileShardResult> and
helps to serialize / ToXContent
- InternalProfileShardResult (singular) is now the holder for all shard-level profiling details. Currently
this includes query, collectors and rewrite. In the future it would hold suggest, aggs, etc
- When the user requests the profiled results, they get back a Map<String, ProfileShardResult>
instead of doing silly helper methods to convert to maps, etc
- Shard details are baked into a string instead of serializing the object
commit 24819ad094b208d0e94f17ce9c3f7c92f7414124
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 10:25:38 2015 -0400
Make Profile results immutable by removing relative_time
commit bfaf095f45fed74194ef78160a8e5dcae1850f9e
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 23 10:54:59 2015 +0200
Add nocommits.
commit e9a128d0d26d5b383b52135ca886f2c987850179
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 23 10:39:37 2015 +0200
Move all profile-related classes to the same package.
commit f20b7c7fdf85384ecc37701bb65310fb8c20844f
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 23 10:33:14 2015 +0200
Reorganize code a bit to ease unit testing of ProfileCollector.
commit 3261306edad6a0c70f59eaee8fe58560f61a75fd
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 18:07:28 2015 +0200
Remove irrelevant nocommit.
commit a6ac868dad12a2e17929878681f66dbd0948d322
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 18:06:45 2015 +0200
Make CollectorResult's reason a free-text field to ease bw compat.
commit 5d0bf170781a950d08b81871cd1e403e49f3cc12
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 16:50:52 2015 +0200
Add unit tests for ProfileWeight/ProfileScorer.
commit 2cd88c412c6e62252504ef69a59216adbb574ce4
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 15:55:17 2015 +0200
Rename InternalQueryProfiler to Profiler.
commit 84f5718fa6779f710da129d9e0e6ff914fd85e36
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 15:53:58 2015 +0200
Merge InternalProfileBreakdown into ProfileBreakdown.
commit 135168eaeb8999c8117ea25288104b0961ce9b35
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 13:56:57 2015 +0200
Make it possible to instantiate a ContextIndexSearcher without SearchContext.
commit 5493fb52376b48460c4ce2dedbe00cc5f6620499
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 11:53:29 2015 +0200
Move ProfileWeight/Scorer to their own files.
commit bf2d917b9dae3b32dfc29c35a7cac4ccb7556cce
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 11:38:24 2015 +0200
Fix bug that caused phrase queries to fail.
commit b2bb0c92c343334ec1703a221af24a1b55e36d53
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 11:36:17 2015 +0200
Parsing happens on the coordinating node now.
commit 416cabb8621acb5cd8dfa77374fd23e428f52fe9
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 11:22:17 2015 +0200
Fix compilation (in particular remove guava deps).
commit f996508645f842629d403fc2e71c1890c0e2cac9
Merge: 4616a25 bc3b91e
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 10:44:38 2015 +0200
Merge branch 'master' into query_profiler
commit 4616a25afffe9c24c6531028f7fccca4303d2893
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Oct 20 12:11:32 2015 -0400
Make Java Count API compatible with profiling
commit cbfba74e16083d719722500ac226efdb5cb2ff55
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Oct 20 12:11:19 2015 -0400
Fix serialization of profile query param, NPE
commit e33ffac383b03247046913da78c8a27e457fae78
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Oct 20 11:17:48 2015 -0400
TestSearchContext should return null Profiler instead of exception
commit 73a02d69b466dc1a5b8a5f022464d6c99e6c2ac3
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Oct 19 12:07:29 2015 -0400
[DOCS] Update docs to reflect new ID format
commit 36248e388c354f954349ecd498db7b66f84ce813
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Oct 19 12:03:03 2015 -0400
Use the full [node][index][shard] string as profile result ID
commit 5cfcc4a6a6b0bcd6ebaa7c8a2d0acc32529a80e1
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 15 17:51:40 2015 -0400
Add failing test for phrase matching
Stack trace generated:
[2015-10-15 17:50:54,438][ERROR][org.elasticsearch.search.profile] shard [[JNj7RX_oSJikcnX72aGBoA][test][2]], reason [RemoteTransportException[[node_s0][local[1]][indices:data/read/search[phase/query]]]; nested: QueryPhaseExecutionException[Query Failed [Failed to execute main query]]; nested: AssertionError[nextPosition() called more than freq() times!]; ], cause [java.lang.AssertionError: nextPosition() called more than freq() times!
at org.apache.lucene.index.AssertingLeafReader$AssertingPostingsEnum.nextPosition(AssertingLeafReader.java:353)
at org.apache.lucene.search.ExactPhraseScorer.phraseFreq(ExactPhraseScorer.java:132)
at org.apache.lucene.search.ExactPhraseScorer.access$000(ExactPhraseScorer.java:27)
at org.apache.lucene.search.ExactPhraseScorer$1.matches(ExactPhraseScorer.java:69)
at org.elasticsearch.common.lucene.search.ProfileQuery$ProfileScorer$2.matches(ProfileQuery.java:226)
at org.apache.lucene.search.ConjunctionDISI$TwoPhaseConjunctionDISI.matches(ConjunctionDISI.java:175)
at org.apache.lucene.search.ConjunctionDISI$TwoPhase.matches(ConjunctionDISI.java:213)
at org.apache.lucene.search.ConjunctionDISI.doNext(ConjunctionDISI.java:128)
at org.apache.lucene.search.ConjunctionDISI.nextDoc(ConjunctionDISI.java:151)
at org.apache.lucene.search.ConjunctionScorer.nextDoc(ConjunctionScorer.java:62)
at org.elasticsearch.common.lucene.search.ProfileQuery$ProfileScorer$1.nextDoc(ProfileQuery.java:205)
at org.apache.lucene.search.Weight$DefaultBulkScorer.scoreAll(Weight.java:224)
at org.apache.lucene.search.Weight$DefaultBulkScorer.score(Weight.java:169)
at org.apache.lucene.search.BulkScorer.score(BulkScorer.java:39)
at org.apache.lucene.search.IndexSearcher.search(IndexSearcher.java:795)
at org.apache.lucene.search.IndexSearcher.search(IndexSearcher.java:509)
at org.elasticsearch.search.query.QueryPhase.execute(QueryPhase.java:347)
at org.elasticsearch.search.query.QueryPhase.execute(QueryPhase.java:111)
at org.elasticsearch.search.SearchService.loadOrExecuteQueryPhase(SearchService.java:366)
at org.elasticsearch.search.SearchService.executeQueryPhase(SearchService.java:378)
at org.elasticsearch.search.action.SearchServiceTransportAction$SearchQueryTransportHandler.messageReceived(SearchServiceTransportAction.java:368)
at org.elasticsearch.search.action.SearchServiceTransportAction$SearchQueryTransportHandler.messageReceived(SearchServiceTransportAction.java:365)
at org.elasticsearch.transport.local.LocalTransport$2.doRun(LocalTransport.java:280)
at org.elasticsearch.common.util.concurrent.AbstractRunnable.run(AbstractRunnable.java:37)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
at java.lang.Thread.run(Thread.java:745)
commit 889fe6383370fe919aaa9f0af398e3040209e40b
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 15 17:30:38 2015 -0400
[DOCS] More docs
commit 89177965d031d84937753538b88ea5ebae2956b0
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 15 09:59:09 2015 -0400
Fix multi-stage rewrites to recursively find most appropriate descendant rewrite
Previously, we chose the first rewrite that matched. But in situations where a query may
rewrite several times, this won't build the tree correctly. Instead we need to recurse
down all the rewrites until we find the most appropriate "leaf" rewrite
The implementation of this is kinda gross: we recursively call getRewrittenParentToken(),
which does a linear scan over the rewriteMap and tries to find rewrites with a larger token
value (since we know child tokens are always larger). Can almost certainly find a better
way to do this...
commit 0b4d782b5348e5d03fd26f7d91bc4a3fbcb7f6a5
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Oct 14 19:30:06 2015 -0400
[Docs] Documentation checkpoint
commit 383636453f6610fcfef9070c21ae7ca11346793e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Sep 16 16:02:22 2015 -0400
Comments
commit a81e8f31e681be16e89ceab9ba3c3e0a018f18ef
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Sep 16 15:48:49 2015 -0400
[TESTS] Ensure all tests use QUERY_THEN_FETCH, DFS does not profile
commit 1255c2d790d85fcb9cbb78bf2a53195138c6bc24
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Sep 15 16:43:46 2015 -0400
Refactor rewrite handling to handle identical rewrites
commit 85b7ec82eb0b26a6fe87266b38f5f86f9ac0c44f
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Sep 15 08:51:14 2015 -0400
Don't update parent when a token is added as root -- Fixes NPE
commit 109d02bdbc49741a3b61e8624521669b0968b839
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Sep 15 08:50:40 2015 -0400
Don't set the rewritten query if not profiling -- Fixes NPE
commit 233cf5e85f6f2c39ed0a2a33d7edd3bbd40856e8
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Sep 14 18:04:51 2015 -0400
Update tests to new response format
commit a930b1fc19de3a329abc8ffddc6711c1246a4b15
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Sep 14 18:03:58 2015 -0400
Fix serialization
commit 69afdd303660510c597df9bada5531b19d134f3d
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Sep 14 15:11:31 2015 -0400
Comments and cleanup
commit 64e7ca7f78187875378382ec5d5aa2462ff71df5
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Sep 14 14:40:21 2015 -0400
Move timing into dedicated class, add proper rewrite integration
commit b44ff85ddbba0a080e65f2e7cc8c50d30e95df8e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Sep 14 12:00:38 2015 -0400
Checkpoint - Refactoring to use a token-based dependency tree
commit 52cedd5266d6a87445c6a4cff3be8ff2087cd1b7
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Sep 4 19:18:19 2015 -0400
Need to set context profiling flag before calling queryPhase.preProcess
commit c524670cb1ce29b4b3a531fa2bff0c403b756f46
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Sep 4 18:00:37 2015 +0200
Reduce profiling overhead a bit.
This removes hash-table lookups everytime we start/stop a profiling clock.
commit 111444ff8418737082236492b37321fc96041e09
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Sep 4 16:18:59 2015 +0200
Add profiling of two-phase iterators.
This is useful for eg. phrase queries or script filters, since they are
typically consumed through their two-phase iterator instead of the scorer.
commit f275e690459e73211bc8494c6de595c0320f4c0b
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Sep 4 16:03:21 2015 +0200
Some more improvements.
I changed profiling to disable bulk scoring, since it makes it impossible to
know where time is spent. Also I removed profiling of operations that are
always fast (eg. normalization) and added nextDoc/advance.
commit 3c8dcd872744de8fd76ce13b6f18f36f8de44068
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Sep 4 14:39:50 2015 +0200
Remove println.
commit d68304862fb38a3823aebed35a263bd9e2176c2f
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Sep 4 14:36:03 2015 +0200
Fix some test failures introduced by the rebase...
commit 04d53ca89fb34b7a21515d770c32aaffcc513b90
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Sep 4 13:57:35 2015 +0200
Reconcile conflicting changes after rebase
commit fed03ec8e2989a0678685cd6c50a566cec42ea4f
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Aug 20 22:40:39 2015 -0400
Add Collectors to profile results
Profile response element has now been re-arranged so that everything is listed per-shard to
facilitate grouping elements together. The new `collector` element looks like this:
```
"profile": {
"shards": [
{
"shard_id": "keP4YFywSXWALCl4m4k24Q",
"query": [...],
"collector": [
{
"name": "MultiCollector",
"purpose": "search_multi",
"time": "16.44504400ms",
"relative_time": "100.0000000%",
"children": [
{
"name": "FilteredCollector",
"purpose": "search_post_filter",
"time": "4.556013000ms",
"relative_time": "27.70447437%",
"children": [
{
"name": "SimpleTopScoreDocCollector",
"purpose": "search_sorted",
"time": "1.352166000ms",
"relative_time": "8.222331299%",
"children": []
}
]
},
{
"name": "BucketCollector: [[non_global_term, another_agg]]",
"purpose": "aggregation",
"time": "10.40379400ms",
"relative_time": "63.26400829%",
"children": []
},
...
```
commit 1368b495c934be642c00f6cbf9fc875d7e6c07ff
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Aug 19 12:43:03 2015 -0400
Move InternalProfiler to profile package
commit 53584de910db6d4a6bb374c9ebb954f204882996
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 18:34:58 2015 -0400
Only reconcile rewrite timing when rewritten query is different from original
commit 9804c3b29d2107cd97f1c7e34d77171b62cb33d0
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 16:40:15 2015 -0400
Comments and cleanup
commit 8e898cc7c59c0c1cc5ed576dfed8e3034ca0967f
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 14:19:07 2015 -0400
[TESTS] Fix comparison test to ensure results sort identically
commit f402a29001933eef29d5a62e81c8563f1c8d0969
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 14:17:59 2015 -0400
Add note about DFS being unable to profile
commit d446e08d3bc91cd85b24fc908e2d82fc5739d598
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 14:17:23 2015 -0400
Implement some missing methods
commit 13ca94fb86fb037a30d181b73d9296153a63d6e4
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 13:10:54 2015 -0400
[TESTS] Comments & cleanup
commit c76c8c771fdeee807761c25938a642612a6ed8e7
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 13:06:08 2015 -0400
[TESTS] Fix profileMatchesRegular to handle NaN scores and nearlyEqual floats
commit 7e7a10ecd26677b2239149468e24938ce5cc18e1
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 12:22:16 2015 -0400
Move nearlyEquals() utility function to shared location
commit 842222900095df4b27ff3593dbb55a42549f2697
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 12:04:35 2015 -0400
Fixup rebase conflicts
commit 674f162d7704dd2034b8361358decdefce1f76ce
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Aug 17 15:29:35 2015 -0400
[TESTS] Update match and bool tests
commit 520380a85456d7137734aed0b06a740e18c9cdec
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Aug 17 15:28:09 2015 -0400
Make naming consistent re: plural
commit b9221501d839bb24d6db575d08e9bee34043fc65
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Aug 17 15:27:39 2015 -0400
Children need to be added to list after serialization
commit 05fa51df940c332fbc140517ee56e849f2d40a72
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Aug 17 15:22:41 2015 -0400
Re-enable bypass for non-profiled queries
commit f132204d264af77a75bd26a02d4e251a19eb411d
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Aug 17 15:21:14 2015 -0400
Fix serialization of QuerySearchResult, InternalProfileResult
commit 27b98fd475fc2e9508c91436ef30624bdbee54ba
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Aug 10 17:39:17 2015 -0400
Start to add back tests, refactor Java api
commit bcfc9fefd49307045108408dc160774666510e85
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 4 17:08:10 2015 -0400
Checkpoint
commit 26a530e0101ce252450eb23e746e48c2fd1bfcae
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Jul 14 13:30:32 2015 -0400
Add createWeight() checkpoint
commit f0dd61de809c5c13682aa213c0be65972537a0df
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Jul 13 12:36:27 2015 -0400
checkpoint
commit 377ee8ce5729b8d388c4719913b48fae77a16686
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Mar 18 10:45:01 2015 -0400
checkpoint
2015-12-17 12:05:43 -05:00
|
|
|
private Profilers profilers;
|
2014-10-13 09:41:55 +02:00
|
|
|
|
2016-09-09 15:28:12 +02:00
|
|
|
private final Map<String, SearchExtBuilder> searchExtBuilders = new HashMap<>();
|
2015-08-14 13:29:33 +02:00
|
|
|
private final Map<Class<?>, Collector> queryCollectors = new HashMap<>();
|
2016-02-08 12:09:15 +01:00
|
|
|
private final QueryShardContext queryShardContext;
|
2018-11-07 20:23:35 +01:00
|
|
|
private final FetchPhase fetchPhase;
|
2015-07-22 17:16:15 +02:00
|
|
|
|
2020-03-09 11:59:30 -04:00
|
|
|
DefaultSearchContext(SearchContextId id, ShardSearchRequest request, SearchShardTarget shardTarget,
|
2018-04-26 09:58:17 +02:00
|
|
|
Engine.Searcher engineSearcher, ClusterService clusterService, IndexService indexService,
|
2019-02-21 12:42:54 +02:00
|
|
|
IndexShard indexShard, BigArrays bigArrays, LongSupplier relativeTimeSupplier, TimeValue timeout,
|
2020-03-23 21:30:34 +01:00
|
|
|
FetchPhase fetchPhase, boolean lowLevelCancellation, Version minNodeVersion) throws IOException {
|
2013-08-26 16:28:27 +02:00
|
|
|
this.id = id;
|
|
|
|
this.request = request;
|
2015-12-16 10:22:54 +00:00
|
|
|
this.fetchPhase = fetchPhase;
|
2013-08-26 16:28:27 +02:00
|
|
|
this.searchType = request.searchType();
|
|
|
|
this.shardTarget = shardTarget;
|
|
|
|
this.engineSearcher = engineSearcher;
|
2014-05-21 13:37:03 +02:00
|
|
|
// SearchContexts use a BigArrays that can circuit break
|
|
|
|
this.bigArrays = bigArrays.withCircuitBreaking();
|
2013-08-26 16:28:27 +02:00
|
|
|
this.dfsResult = new DfsSearchResult(id, shardTarget);
|
|
|
|
this.queryResult = new QuerySearchResult(id, shardTarget);
|
|
|
|
this.fetchResult = new FetchSearchResult(id, shardTarget);
|
|
|
|
this.indexShard = indexShard;
|
|
|
|
this.indexService = indexService;
|
2018-04-26 09:58:17 +02:00
|
|
|
this.clusterService = clusterService;
|
2019-07-04 22:48:48 +02:00
|
|
|
this.searcher = new ContextIndexSearcher(engineSearcher.getIndexReader(), engineSearcher.getSimilarity(),
|
2020-03-23 21:30:34 +01:00
|
|
|
engineSearcher.getQueryCache(), engineSearcher.getQueryCachingPolicy(), lowLevelCancellation);
|
2019-02-21 12:42:54 +02:00
|
|
|
this.relativeTimeSupplier = relativeTimeSupplier;
|
2016-06-27 18:41:18 -04:00
|
|
|
this.timeout = timeout;
|
2018-04-26 09:58:17 +02:00
|
|
|
this.minNodeVersion = minNodeVersion;
|
2019-09-11 16:12:34 +02:00
|
|
|
queryShardContext = indexService.newQueryShardContext(request.shardId().id(), searcher,
|
|
|
|
request::nowInMillis, shardTarget.getClusterAlias());
|
2016-02-08 12:09:15 +01:00
|
|
|
queryShardContext.setTypes(request.types());
|
2016-11-07 17:20:06 +09:00
|
|
|
queryBoost = request.indexBoost();
|
2020-03-23 21:30:34 +01:00
|
|
|
this.lowLevelCancellation = lowLevelCancellation;
|
2013-08-26 16:28:27 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
@Override
|
2015-04-29 10:08:50 +02:00
|
|
|
public void doClose() {
|
2019-07-04 22:48:48 +02:00
|
|
|
Releasables.close(engineSearcher);
|
2013-09-30 22:50:29 +02:00
|
|
|
}
|
|
|
|
|
2013-08-26 16:28:27 +02:00
|
|
|
/**
|
|
|
|
* Should be called before executing the main query and after all other parameters have been set.
|
|
|
|
*/
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2016-09-12 22:42:55 +02:00
|
|
|
public void preProcess(boolean rewrite) {
|
2016-03-18 14:02:34 -04:00
|
|
|
if (hasOnlySuggest() ) {
|
|
|
|
return;
|
|
|
|
}
|
2016-07-11 12:34:58 -04:00
|
|
|
long from = from() == -1 ? 0 : from();
|
|
|
|
long size = size() == -1 ? 10 : size();
|
|
|
|
long resultWindow = from + size;
|
|
|
|
int maxResultWindow = indexService.getIndexSettings().getMaxResultWindow();
|
2015-08-28 14:16:45 -04:00
|
|
|
|
2016-07-11 12:34:58 -04:00
|
|
|
if (resultWindow > maxResultWindow) {
|
|
|
|
if (scrollContext == null) {
|
2017-10-31 06:17:27 -05:00
|
|
|
throw new IllegalArgumentException(
|
2015-08-28 14:16:45 -04:00
|
|
|
"Result window is too large, from + size must be less than or equal to: [" + maxResultWindow + "] but was ["
|
|
|
|
+ resultWindow + "]. See the scroll api for a more efficient way to request large data sets. "
|
2016-01-14 15:21:59 +01:00
|
|
|
+ "This limit can be set by changing the [" + IndexSettings.MAX_RESULT_WINDOW_SETTING.getKey()
|
2016-04-21 12:59:19 -04:00
|
|
|
+ "] index level setting.");
|
|
|
|
}
|
2017-10-31 06:17:27 -05:00
|
|
|
throw new IllegalArgumentException(
|
2016-07-11 12:34:58 -04:00
|
|
|
"Batch size is too large, size must be less than or equal to: [" + maxResultWindow + "] but was [" + resultWindow
|
|
|
|
+ "]. Scroll batch sizes cost as much memory as result windows so they are controlled by the ["
|
|
|
|
+ IndexSettings.MAX_RESULT_WINDOW_SETTING.getKey() + "] index level setting.");
|
2016-04-21 12:59:19 -04:00
|
|
|
}
|
|
|
|
if (rescore != null) {
|
2017-09-07 14:17:37 +02:00
|
|
|
if (sort != null) {
|
2017-10-31 06:17:27 -05:00
|
|
|
throw new IllegalArgumentException("Cannot use [sort] option in conjunction with [rescore].");
|
2017-09-07 14:17:37 +02:00
|
|
|
}
|
2016-04-21 12:59:19 -04:00
|
|
|
int maxWindow = indexService.getIndexSettings().getMaxRescoreWindow();
|
2017-08-25 13:46:57 -04:00
|
|
|
for (RescoreContext rescoreContext: rescore) {
|
|
|
|
if (rescoreContext.getWindowSize() > maxWindow) {
|
2017-10-31 06:17:27 -05:00
|
|
|
throw new IllegalArgumentException("Rescore window [" + rescoreContext.getWindowSize() + "] is too large. "
|
2017-08-25 13:46:57 -04:00
|
|
|
+ "It must be less than [" + maxWindow + "]. This prevents allocating massive heaps for storing the results "
|
|
|
|
+ "to be rescored. This limit can be set by changing the [" + IndexSettings.MAX_RESCORE_WINDOW_SETTING.getKey()
|
2016-04-21 12:59:19 -04:00
|
|
|
+ "] index level setting.");
|
|
|
|
}
|
2014-09-18 12:25:48 +02:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2016-06-08 15:28:06 +02:00
|
|
|
if (sliceBuilder != null) {
|
|
|
|
int sliceLimit = indexService.getIndexSettings().getMaxSlicesPerScroll();
|
|
|
|
int numSlices = sliceBuilder.getMax();
|
|
|
|
if (numSlices > sliceLimit) {
|
2017-10-31 06:17:27 -05:00
|
|
|
throw new IllegalArgumentException("The number of slices [" + numSlices + "] is too large. It must "
|
2016-06-08 15:28:06 +02:00
|
|
|
+ "be less than [" + sliceLimit + "]. This limit can be set by changing the [" +
|
|
|
|
IndexSettings.MAX_SLICES_PER_SCROLL.getKey() + "] index level setting.");
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2015-06-30 01:11:27 +02:00
|
|
|
// initialize the filtering alias based on the provided filters
|
2016-10-14 16:26:09 +02:00
|
|
|
try {
|
2017-07-20 15:37:50 +02:00
|
|
|
final QueryBuilder queryBuilder = request.getAliasFilter().getQueryBuilder();
|
2018-12-03 11:49:11 +01:00
|
|
|
aliasFilter = queryBuilder == null ? null : queryBuilder.toQuery(queryShardContext);
|
2016-10-14 16:26:09 +02:00
|
|
|
} catch (IOException e) {
|
|
|
|
throw new UncheckedIOException(e);
|
|
|
|
}
|
2015-06-30 01:11:27 +02:00
|
|
|
|
2013-08-26 16:28:27 +02:00
|
|
|
if (query() == null) {
|
|
|
|
parsedQuery(ParsedQuery.parsedMatchAllQuery());
|
|
|
|
}
|
2015-10-23 11:55:28 +02:00
|
|
|
if (queryBoost() != AbstractQueryBuilder.DEFAULT_BOOST) {
|
2020-02-24 14:28:51 +00:00
|
|
|
parsedQuery(new ParsedQuery(new BoostQuery(query(), queryBoost), parsedQuery()));
|
2013-08-26 16:28:27 +02:00
|
|
|
}
|
Nested queries should avoid adding unnecessary filters when possible. (#23079)
When nested objects are present in the mappings, many queries get deoptimized
due to the need to exclude documents that are not in the right space. For
instance, a filter is applied to all queries that prevents them from matching
non-root documents (`+*:* -_type:__*`). Moreover, a filter is applied to all
child queries of `nested` queries in order to make sure that the child query
only matches child documents (`_type:__nested_path`), which is required by
`ToParentBlockJoinQuery` (the Lucene query behing Elasticsearch's `nested`
queries).
These additional filters slow down `nested` queries. In 1.7-, the cost was
somehow amortized by the fact that we cached filters very aggressively. However,
this has proven to be a significant source of slow downs since 2.0 for users
of `nested` mappings and queries, see #20797.
This change makes the filtering a bit smarter. For instance if the query is a
`match_all` query, then we need to exclude nested docs. However, if the query
is `foo: bar` then it may only match root documents since `foo` is a top-level
field, so no additional filtering is required.
Another improvement is to use a `FILTER` clause on all types rather than a
`MUST_NOT` clause on all nested paths when possible since `FILTER` clauses
are more efficient.
Here are some examples of queries and how they get rewritten:
```
"match_all": {}
```
This query gets rewritten to `ConstantScore(+*:* -_type:__*)` on master and
`ConstantScore(_type:AutomatonQuery {\norg.apache.lucene.util.automaton.Automaton@4371da44})`
with this change. The automaton is the complement of `_type:__*` so it matches
the same documents, but is faster since it is now a positive clause. Simplistic
performance testing on a 10M index where each root document has 5 nested
documents on average gave a latency of 420ms on master and 90ms with this change
applied.
```
"term": {
"foo": {
"value": "0"
}
}
```
This query is rewritten to `+foo:0 #(ConstantScore(+*:* -_type:__*))^0.0` on
master and `foo:0` with this change: we do not need to filter nested docs out
since the query cannot match nested docs. While doing performance testing in
the same conditions as above, response times went from 250ms to 50ms.
```
"nested": {
"path": "nested",
"query": {
"term": {
"nested.foo": {
"value": "0"
}
}
}
}
```
This query is rewritten to
`+ToParentBlockJoinQuery (+nested.foo:0 #_type:__nested) #(ConstantScore(+*:* -_type:__*))^0.0`
on master and `ToParentBlockJoinQuery (nested.foo:0)` with this change. The
top-level filter (`-_type:__*`) could be removed since `nested` queries only
match documents of the parent space, as well as the child filter
(`#_type:__nested`) since the child query may only match nested docs since the
`nested` object has both `include_in_parent` and `include_in_root` set to
`false`. While doing performance testing in the same conditions as above,
response times went from 850ms to 270ms.
2017-02-14 16:05:19 +01:00
|
|
|
this.query = buildFilteredQuery(query);
|
2016-09-12 22:42:55 +02:00
|
|
|
if (rewrite) {
|
|
|
|
try {
|
|
|
|
this.query = searcher.rewrite(query);
|
|
|
|
} catch (IOException e) {
|
2019-09-26 14:21:23 +02:00
|
|
|
throw new QueryPhaseExecutionException(shardTarget, "Failed to rewrite main query", e);
|
2016-09-12 22:42:55 +02:00
|
|
|
}
|
2015-08-14 13:29:33 +02:00
|
|
|
}
|
2013-08-26 16:28:27 +02:00
|
|
|
}
|
|
|
|
|
Nested queries should avoid adding unnecessary filters when possible. (#23079)
When nested objects are present in the mappings, many queries get deoptimized
due to the need to exclude documents that are not in the right space. For
instance, a filter is applied to all queries that prevents them from matching
non-root documents (`+*:* -_type:__*`). Moreover, a filter is applied to all
child queries of `nested` queries in order to make sure that the child query
only matches child documents (`_type:__nested_path`), which is required by
`ToParentBlockJoinQuery` (the Lucene query behing Elasticsearch's `nested`
queries).
These additional filters slow down `nested` queries. In 1.7-, the cost was
somehow amortized by the fact that we cached filters very aggressively. However,
this has proven to be a significant source of slow downs since 2.0 for users
of `nested` mappings and queries, see #20797.
This change makes the filtering a bit smarter. For instance if the query is a
`match_all` query, then we need to exclude nested docs. However, if the query
is `foo: bar` then it may only match root documents since `foo` is a top-level
field, so no additional filtering is required.
Another improvement is to use a `FILTER` clause on all types rather than a
`MUST_NOT` clause on all nested paths when possible since `FILTER` clauses
are more efficient.
Here are some examples of queries and how they get rewritten:
```
"match_all": {}
```
This query gets rewritten to `ConstantScore(+*:* -_type:__*)` on master and
`ConstantScore(_type:AutomatonQuery {\norg.apache.lucene.util.automaton.Automaton@4371da44})`
with this change. The automaton is the complement of `_type:__*` so it matches
the same documents, but is faster since it is now a positive clause. Simplistic
performance testing on a 10M index where each root document has 5 nested
documents on average gave a latency of 420ms on master and 90ms with this change
applied.
```
"term": {
"foo": {
"value": "0"
}
}
```
This query is rewritten to `+foo:0 #(ConstantScore(+*:* -_type:__*))^0.0` on
master and `foo:0` with this change: we do not need to filter nested docs out
since the query cannot match nested docs. While doing performance testing in
the same conditions as above, response times went from 250ms to 50ms.
```
"nested": {
"path": "nested",
"query": {
"term": {
"nested.foo": {
"value": "0"
}
}
}
}
```
This query is rewritten to
`+ToParentBlockJoinQuery (+nested.foo:0 #_type:__nested) #(ConstantScore(+*:* -_type:__*))^0.0`
on master and `ToParentBlockJoinQuery (nested.foo:0)` with this change. The
top-level filter (`-_type:__*`) could be removed since `nested` queries only
match documents of the parent space, as well as the child filter
(`#_type:__nested`) since the child query may only match nested docs since the
`nested` object has both `include_in_parent` and `include_in_root` set to
`false`. While doing performance testing in the same conditions as above,
response times went from 850ms to 270ms.
2017-02-14 16:05:19 +01:00
|
|
|
@Override
|
|
|
|
public Query buildFilteredQuery(Query query) {
|
|
|
|
List<Query> filters = new ArrayList<>();
|
|
|
|
Query typeFilter = createTypeFilter(queryShardContext.getTypes());
|
|
|
|
if (typeFilter != null) {
|
|
|
|
filters.add(typeFilter);
|
2016-01-06 12:12:20 -05:00
|
|
|
}
|
Nested queries should avoid adding unnecessary filters when possible. (#23079)
When nested objects are present in the mappings, many queries get deoptimized
due to the need to exclude documents that are not in the right space. For
instance, a filter is applied to all queries that prevents them from matching
non-root documents (`+*:* -_type:__*`). Moreover, a filter is applied to all
child queries of `nested` queries in order to make sure that the child query
only matches child documents (`_type:__nested_path`), which is required by
`ToParentBlockJoinQuery` (the Lucene query behing Elasticsearch's `nested`
queries).
These additional filters slow down `nested` queries. In 1.7-, the cost was
somehow amortized by the fact that we cached filters very aggressively. However,
this has proven to be a significant source of slow downs since 2.0 for users
of `nested` mappings and queries, see #20797.
This change makes the filtering a bit smarter. For instance if the query is a
`match_all` query, then we need to exclude nested docs. However, if the query
is `foo: bar` then it may only match root documents since `foo` is a top-level
field, so no additional filtering is required.
Another improvement is to use a `FILTER` clause on all types rather than a
`MUST_NOT` clause on all nested paths when possible since `FILTER` clauses
are more efficient.
Here are some examples of queries and how they get rewritten:
```
"match_all": {}
```
This query gets rewritten to `ConstantScore(+*:* -_type:__*)` on master and
`ConstantScore(_type:AutomatonQuery {\norg.apache.lucene.util.automaton.Automaton@4371da44})`
with this change. The automaton is the complement of `_type:__*` so it matches
the same documents, but is faster since it is now a positive clause. Simplistic
performance testing on a 10M index where each root document has 5 nested
documents on average gave a latency of 420ms on master and 90ms with this change
applied.
```
"term": {
"foo": {
"value": "0"
}
}
```
This query is rewritten to `+foo:0 #(ConstantScore(+*:* -_type:__*))^0.0` on
master and `foo:0` with this change: we do not need to filter nested docs out
since the query cannot match nested docs. While doing performance testing in
the same conditions as above, response times went from 250ms to 50ms.
```
"nested": {
"path": "nested",
"query": {
"term": {
"nested.foo": {
"value": "0"
}
}
}
}
```
This query is rewritten to
`+ToParentBlockJoinQuery (+nested.foo:0 #_type:__nested) #(ConstantScore(+*:* -_type:__*))^0.0`
on master and `ToParentBlockJoinQuery (nested.foo:0)` with this change. The
top-level filter (`-_type:__*`) could be removed since `nested` queries only
match documents of the parent space, as well as the child filter
(`#_type:__nested`) since the child query may only match nested docs since the
`nested` object has both `include_in_parent` and `include_in_root` set to
`false`. While doing performance testing in the same conditions as above,
response times went from 850ms to 270ms.
2017-02-14 16:05:19 +01:00
|
|
|
|
|
|
|
if (mapperService().hasNested()
|
|
|
|
&& typeFilter == null // when a _type filter is set, it will automatically exclude nested docs
|
|
|
|
&& new NestedHelper(mapperService()).mightMatchNestedDocs(query)
|
|
|
|
&& (aliasFilter == null || new NestedHelper(mapperService()).mightMatchNestedDocs(aliasFilter))) {
|
2017-11-21 15:14:03 +01:00
|
|
|
filters.add(Queries.newNonNestedFilter(mapperService().getIndexSettings().getIndexVersionCreated()));
|
2016-01-06 12:12:20 -05:00
|
|
|
}
|
|
|
|
|
Nested queries should avoid adding unnecessary filters when possible. (#23079)
When nested objects are present in the mappings, many queries get deoptimized
due to the need to exclude documents that are not in the right space. For
instance, a filter is applied to all queries that prevents them from matching
non-root documents (`+*:* -_type:__*`). Moreover, a filter is applied to all
child queries of `nested` queries in order to make sure that the child query
only matches child documents (`_type:__nested_path`), which is required by
`ToParentBlockJoinQuery` (the Lucene query behing Elasticsearch's `nested`
queries).
These additional filters slow down `nested` queries. In 1.7-, the cost was
somehow amortized by the fact that we cached filters very aggressively. However,
this has proven to be a significant source of slow downs since 2.0 for users
of `nested` mappings and queries, see #20797.
This change makes the filtering a bit smarter. For instance if the query is a
`match_all` query, then we need to exclude nested docs. However, if the query
is `foo: bar` then it may only match root documents since `foo` is a top-level
field, so no additional filtering is required.
Another improvement is to use a `FILTER` clause on all types rather than a
`MUST_NOT` clause on all nested paths when possible since `FILTER` clauses
are more efficient.
Here are some examples of queries and how they get rewritten:
```
"match_all": {}
```
This query gets rewritten to `ConstantScore(+*:* -_type:__*)` on master and
`ConstantScore(_type:AutomatonQuery {\norg.apache.lucene.util.automaton.Automaton@4371da44})`
with this change. The automaton is the complement of `_type:__*` so it matches
the same documents, but is faster since it is now a positive clause. Simplistic
performance testing on a 10M index where each root document has 5 nested
documents on average gave a latency of 420ms on master and 90ms with this change
applied.
```
"term": {
"foo": {
"value": "0"
}
}
```
This query is rewritten to `+foo:0 #(ConstantScore(+*:* -_type:__*))^0.0` on
master and `foo:0` with this change: we do not need to filter nested docs out
since the query cannot match nested docs. While doing performance testing in
the same conditions as above, response times went from 250ms to 50ms.
```
"nested": {
"path": "nested",
"query": {
"term": {
"nested.foo": {
"value": "0"
}
}
}
}
```
This query is rewritten to
`+ToParentBlockJoinQuery (+nested.foo:0 #_type:__nested) #(ConstantScore(+*:* -_type:__*))^0.0`
on master and `ToParentBlockJoinQuery (nested.foo:0)` with this change. The
top-level filter (`-_type:__*`) could be removed since `nested` queries only
match documents of the parent space, as well as the child filter
(`#_type:__nested`) since the child query may only match nested docs since the
`nested` object has both `include_in_parent` and `include_in_root` set to
`false`. While doing performance testing in the same conditions as above,
response times went from 850ms to 270ms.
2017-02-14 16:05:19 +01:00
|
|
|
if (aliasFilter != null) {
|
|
|
|
filters.add(aliasFilter);
|
2016-05-10 12:10:55 +02:00
|
|
|
}
|
Nested queries should avoid adding unnecessary filters when possible. (#23079)
When nested objects are present in the mappings, many queries get deoptimized
due to the need to exclude documents that are not in the right space. For
instance, a filter is applied to all queries that prevents them from matching
non-root documents (`+*:* -_type:__*`). Moreover, a filter is applied to all
child queries of `nested` queries in order to make sure that the child query
only matches child documents (`_type:__nested_path`), which is required by
`ToParentBlockJoinQuery` (the Lucene query behing Elasticsearch's `nested`
queries).
These additional filters slow down `nested` queries. In 1.7-, the cost was
somehow amortized by the fact that we cached filters very aggressively. However,
this has proven to be a significant source of slow downs since 2.0 for users
of `nested` mappings and queries, see #20797.
This change makes the filtering a bit smarter. For instance if the query is a
`match_all` query, then we need to exclude nested docs. However, if the query
is `foo: bar` then it may only match root documents since `foo` is a top-level
field, so no additional filtering is required.
Another improvement is to use a `FILTER` clause on all types rather than a
`MUST_NOT` clause on all nested paths when possible since `FILTER` clauses
are more efficient.
Here are some examples of queries and how they get rewritten:
```
"match_all": {}
```
This query gets rewritten to `ConstantScore(+*:* -_type:__*)` on master and
`ConstantScore(_type:AutomatonQuery {\norg.apache.lucene.util.automaton.Automaton@4371da44})`
with this change. The automaton is the complement of `_type:__*` so it matches
the same documents, but is faster since it is now a positive clause. Simplistic
performance testing on a 10M index where each root document has 5 nested
documents on average gave a latency of 420ms on master and 90ms with this change
applied.
```
"term": {
"foo": {
"value": "0"
}
}
```
This query is rewritten to `+foo:0 #(ConstantScore(+*:* -_type:__*))^0.0` on
master and `foo:0` with this change: we do not need to filter nested docs out
since the query cannot match nested docs. While doing performance testing in
the same conditions as above, response times went from 250ms to 50ms.
```
"nested": {
"path": "nested",
"query": {
"term": {
"nested.foo": {
"value": "0"
}
}
}
}
```
This query is rewritten to
`+ToParentBlockJoinQuery (+nested.foo:0 #_type:__nested) #(ConstantScore(+*:* -_type:__*))^0.0`
on master and `ToParentBlockJoinQuery (nested.foo:0)` with this change. The
top-level filter (`-_type:__*`) could be removed since `nested` queries only
match documents of the parent space, as well as the child filter
(`#_type:__nested`) since the child query may only match nested docs since the
`nested` object has both `include_in_parent` and `include_in_root` set to
`false`. While doing performance testing in the same conditions as above,
response times went from 850ms to 270ms.
2017-02-14 16:05:19 +01:00
|
|
|
|
|
|
|
if (sliceBuilder != null) {
|
2020-01-22 11:43:15 -06:00
|
|
|
Query slicedQuery = sliceBuilder.toFilter(clusterService, request, queryShardContext, minNodeVersion);
|
|
|
|
if (slicedQuery instanceof MatchNoDocsQuery) {
|
|
|
|
return slicedQuery;
|
|
|
|
} else {
|
|
|
|
filters.add(slicedQuery);
|
|
|
|
}
|
Nested queries should avoid adding unnecessary filters when possible. (#23079)
When nested objects are present in the mappings, many queries get deoptimized
due to the need to exclude documents that are not in the right space. For
instance, a filter is applied to all queries that prevents them from matching
non-root documents (`+*:* -_type:__*`). Moreover, a filter is applied to all
child queries of `nested` queries in order to make sure that the child query
only matches child documents (`_type:__nested_path`), which is required by
`ToParentBlockJoinQuery` (the Lucene query behing Elasticsearch's `nested`
queries).
These additional filters slow down `nested` queries. In 1.7-, the cost was
somehow amortized by the fact that we cached filters very aggressively. However,
this has proven to be a significant source of slow downs since 2.0 for users
of `nested` mappings and queries, see #20797.
This change makes the filtering a bit smarter. For instance if the query is a
`match_all` query, then we need to exclude nested docs. However, if the query
is `foo: bar` then it may only match root documents since `foo` is a top-level
field, so no additional filtering is required.
Another improvement is to use a `FILTER` clause on all types rather than a
`MUST_NOT` clause on all nested paths when possible since `FILTER` clauses
are more efficient.
Here are some examples of queries and how they get rewritten:
```
"match_all": {}
```
This query gets rewritten to `ConstantScore(+*:* -_type:__*)` on master and
`ConstantScore(_type:AutomatonQuery {\norg.apache.lucene.util.automaton.Automaton@4371da44})`
with this change. The automaton is the complement of `_type:__*` so it matches
the same documents, but is faster since it is now a positive clause. Simplistic
performance testing on a 10M index where each root document has 5 nested
documents on average gave a latency of 420ms on master and 90ms with this change
applied.
```
"term": {
"foo": {
"value": "0"
}
}
```
This query is rewritten to `+foo:0 #(ConstantScore(+*:* -_type:__*))^0.0` on
master and `foo:0` with this change: we do not need to filter nested docs out
since the query cannot match nested docs. While doing performance testing in
the same conditions as above, response times went from 250ms to 50ms.
```
"nested": {
"path": "nested",
"query": {
"term": {
"nested.foo": {
"value": "0"
}
}
}
}
```
This query is rewritten to
`+ToParentBlockJoinQuery (+nested.foo:0 #_type:__nested) #(ConstantScore(+*:* -_type:__*))^0.0`
on master and `ToParentBlockJoinQuery (nested.foo:0)` with this change. The
top-level filter (`-_type:__*`) could be removed since `nested` queries only
match documents of the parent space, as well as the child filter
(`#_type:__nested`) since the child query may only match nested docs since the
`nested` object has both `include_in_parent` and `include_in_root` set to
`false`. While doing performance testing in the same conditions as above,
response times went from 850ms to 270ms.
2017-02-14 16:05:19 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
if (filters.isEmpty()) {
|
|
|
|
return query;
|
|
|
|
} else {
|
|
|
|
BooleanQuery.Builder builder = new BooleanQuery.Builder();
|
|
|
|
builder.add(query, Occur.MUST);
|
|
|
|
for (Query filter : filters) {
|
|
|
|
builder.add(filter, Occur.FILTER);
|
|
|
|
}
|
|
|
|
return builder.build();
|
2016-05-10 12:10:55 +02:00
|
|
|
}
|
2016-01-20 21:06:06 +01:00
|
|
|
}
|
|
|
|
|
2017-05-04 16:29:35 +02:00
|
|
|
private Query createTypeFilter(String[] types) {
|
2016-01-20 21:06:06 +01:00
|
|
|
if (types != null && types.length >= 1) {
|
2020-02-07 10:24:01 -08:00
|
|
|
MappedFieldType ft = mapperService().fieldType(TypeFieldMapper.NAME);
|
2017-05-04 16:29:35 +02:00
|
|
|
if (ft != null) {
|
|
|
|
// ft might be null if no documents have been indexed yet
|
|
|
|
return ft.termsQuery(Arrays.asList(types), queryShardContext);
|
2016-01-20 21:06:06 +01:00
|
|
|
}
|
2016-01-20 23:16:42 +01:00
|
|
|
}
|
Nested queries should avoid adding unnecessary filters when possible. (#23079)
When nested objects are present in the mappings, many queries get deoptimized
due to the need to exclude documents that are not in the right space. For
instance, a filter is applied to all queries that prevents them from matching
non-root documents (`+*:* -_type:__*`). Moreover, a filter is applied to all
child queries of `nested` queries in order to make sure that the child query
only matches child documents (`_type:__nested_path`), which is required by
`ToParentBlockJoinQuery` (the Lucene query behing Elasticsearch's `nested`
queries).
These additional filters slow down `nested` queries. In 1.7-, the cost was
somehow amortized by the fact that we cached filters very aggressively. However,
this has proven to be a significant source of slow downs since 2.0 for users
of `nested` mappings and queries, see #20797.
This change makes the filtering a bit smarter. For instance if the query is a
`match_all` query, then we need to exclude nested docs. However, if the query
is `foo: bar` then it may only match root documents since `foo` is a top-level
field, so no additional filtering is required.
Another improvement is to use a `FILTER` clause on all types rather than a
`MUST_NOT` clause on all nested paths when possible since `FILTER` clauses
are more efficient.
Here are some examples of queries and how they get rewritten:
```
"match_all": {}
```
This query gets rewritten to `ConstantScore(+*:* -_type:__*)` on master and
`ConstantScore(_type:AutomatonQuery {\norg.apache.lucene.util.automaton.Automaton@4371da44})`
with this change. The automaton is the complement of `_type:__*` so it matches
the same documents, but is faster since it is now a positive clause. Simplistic
performance testing on a 10M index where each root document has 5 nested
documents on average gave a latency of 420ms on master and 90ms with this change
applied.
```
"term": {
"foo": {
"value": "0"
}
}
```
This query is rewritten to `+foo:0 #(ConstantScore(+*:* -_type:__*))^0.0` on
master and `foo:0` with this change: we do not need to filter nested docs out
since the query cannot match nested docs. While doing performance testing in
the same conditions as above, response times went from 250ms to 50ms.
```
"nested": {
"path": "nested",
"query": {
"term": {
"nested.foo": {
"value": "0"
}
}
}
}
```
This query is rewritten to
`+ToParentBlockJoinQuery (+nested.foo:0 #_type:__nested) #(ConstantScore(+*:* -_type:__*))^0.0`
on master and `ToParentBlockJoinQuery (nested.foo:0)` with this change. The
top-level filter (`-_type:__*`) could be removed since `nested` queries only
match documents of the parent space, as well as the child filter
(`#_type:__nested`) since the child query may only match nested docs since the
`nested` object has both `include_in_parent` and `include_in_root` set to
`false`. While doing performance testing in the same conditions as above,
response times went from 850ms to 270ms.
2017-02-14 16:05:19 +01:00
|
|
|
return null;
|
2013-08-26 16:28:27 +02:00
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2020-03-09 11:59:30 -04:00
|
|
|
public SearchContextId id() {
|
2013-08-26 16:28:27 +02:00
|
|
|
return this.id;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-09-30 22:50:29 +02:00
|
|
|
public String source() {
|
|
|
|
return engineSearcher.source();
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public ShardSearchRequest request() {
|
|
|
|
return this.request;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public SearchType searchType() {
|
|
|
|
return this.searchType;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public SearchShardTarget shardTarget() {
|
|
|
|
return this.shardTarget;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public int numberOfShards() {
|
|
|
|
return request.numberOfShards();
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public float queryBoost() {
|
|
|
|
return queryBoost;
|
|
|
|
}
|
|
|
|
|
2015-07-20 12:42:13 -04:00
|
|
|
@Override
|
|
|
|
public long getOriginNanoTime() {
|
|
|
|
return originNanoTime;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2015-08-18 17:01:47 +02:00
|
|
|
public ScrollContext scrollContext() {
|
|
|
|
return this.scrollContext;
|
2013-08-26 16:28:27 +02:00
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2015-08-18 17:01:47 +02:00
|
|
|
public SearchContext scrollContext(ScrollContext scrollContext) {
|
|
|
|
this.scrollContext = scrollContext;
|
2013-08-26 16:28:27 +02:00
|
|
|
return this;
|
|
|
|
}
|
|
|
|
|
2013-11-24 03:13:08 -08:00
|
|
|
@Override
|
|
|
|
public SearchContextAggregations aggregations() {
|
|
|
|
return aggregations;
|
|
|
|
}
|
|
|
|
|
|
|
|
@Override
|
|
|
|
public SearchContext aggregations(SearchContextAggregations aggregations) {
|
|
|
|
this.aggregations = aggregations;
|
|
|
|
return this;
|
|
|
|
}
|
|
|
|
|
2015-07-22 17:16:15 +02:00
|
|
|
@Override
|
2016-09-09 15:28:12 +02:00
|
|
|
public void addSearchExt(SearchExtBuilder searchExtBuilder) {
|
|
|
|
//it's ok to use the writeable name here given that we enforce it to be the same as the name of the element that gets
|
|
|
|
//parsed by the corresponding parser. There is one single name and one single way to retrieve the parsed object from the context.
|
|
|
|
searchExtBuilders.put(searchExtBuilder.getWriteableName(), searchExtBuilder);
|
2015-07-22 17:16:15 +02:00
|
|
|
}
|
|
|
|
|
2016-09-07 23:43:17 +02:00
|
|
|
@Override
|
2016-09-09 15:28:12 +02:00
|
|
|
public SearchExtBuilder getSearchExt(String name) {
|
|
|
|
return searchExtBuilders.get(name);
|
2016-09-07 23:43:17 +02:00
|
|
|
}
|
2015-07-29 17:20:06 +02:00
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2020-07-29 11:39:17 -07:00
|
|
|
public SearchHighlightContext highlight() {
|
2013-08-26 16:28:27 +02:00
|
|
|
return highlight;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2020-07-29 11:39:17 -07:00
|
|
|
public void highlight(SearchHighlightContext highlight) {
|
2013-08-26 16:28:27 +02:00
|
|
|
this.highlight = highlight;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public SuggestionSearchContext suggest() {
|
|
|
|
return suggest;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public void suggest(SuggestionSearchContext suggest) {
|
|
|
|
this.suggest = suggest;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2017-08-25 13:46:57 -04:00
|
|
|
public List<RescoreContext> rescore() {
|
2014-01-15 17:58:22 -05:00
|
|
|
if (rescore == null) {
|
|
|
|
return Collections.emptyList();
|
|
|
|
}
|
|
|
|
return rescore;
|
2013-08-26 16:28:27 +02:00
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2017-08-25 13:46:57 -04:00
|
|
|
public void addRescore(RescoreContext rescore) {
|
2014-01-15 17:58:22 -05:00
|
|
|
if (this.rescore == null) {
|
2014-03-27 15:54:45 +01:00
|
|
|
this.rescore = new ArrayList<>();
|
2014-01-15 17:58:22 -05:00
|
|
|
}
|
|
|
|
this.rescore.add(rescore);
|
2013-08-26 16:28:27 +02:00
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public boolean hasScriptFields() {
|
|
|
|
return scriptFields != null;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public ScriptFieldsContext scriptFields() {
|
|
|
|
if (scriptFields == null) {
|
|
|
|
scriptFields = new ScriptFieldsContext();
|
|
|
|
}
|
|
|
|
return this.scriptFields;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
|
|
|
* A shortcut function to see whether there is a fetchSourceContext and it says the source is requested.
|
|
|
|
*/
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public boolean sourceRequested() {
|
|
|
|
return fetchSourceContext != null && fetchSourceContext.fetchSource();
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public boolean hasFetchSourceContext() {
|
|
|
|
return fetchSourceContext != null;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public FetchSourceContext fetchSourceContext() {
|
|
|
|
return this.fetchSourceContext;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public SearchContext fetchSourceContext(FetchSourceContext fetchSourceContext) {
|
|
|
|
this.fetchSourceContext = fetchSourceContext;
|
|
|
|
return this;
|
|
|
|
}
|
|
|
|
|
2016-09-07 17:47:08 +02:00
|
|
|
@Override
|
2020-02-13 12:59:59 -08:00
|
|
|
public FetchDocValuesContext docValuesContext() {
|
|
|
|
return docValuesContext;
|
2016-09-07 17:47:08 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
@Override
|
2020-02-13 12:59:59 -08:00
|
|
|
public SearchContext docValuesContext(FetchDocValuesContext docValuesContext) {
|
|
|
|
this.docValuesContext = docValuesContext;
|
2016-09-07 17:47:08 +02:00
|
|
|
return this;
|
|
|
|
}
|
|
|
|
|
2020-07-28 10:58:20 -07:00
|
|
|
@Override
|
|
|
|
public FetchFieldsContext fetchFieldsContext() {
|
|
|
|
return fetchFieldsContext;
|
|
|
|
}
|
|
|
|
|
|
|
|
@Override
|
|
|
|
public SearchContext fetchFieldsContext(FetchFieldsContext fetchFieldsContext) {
|
|
|
|
this.fetchFieldsContext = fetchFieldsContext;
|
|
|
|
return this;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public ContextIndexSearcher searcher() {
|
|
|
|
return this.searcher;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public IndexShard indexShard() {
|
|
|
|
return this.indexShard;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public MapperService mapperService() {
|
|
|
|
return indexService.mapperService();
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public SimilarityService similarityService() {
|
|
|
|
return indexService.similarityService();
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2014-02-26 16:40:15 +01:00
|
|
|
public BigArrays bigArrays() {
|
|
|
|
return bigArrays;
|
|
|
|
}
|
|
|
|
|
2014-07-25 16:17:48 +02:00
|
|
|
@Override
|
Upgrade master to lucene 5.0 snapshot
This has a lot of improvements in lucene, particularly around memory usage, merging, safety, compressed bitsets, etc.
On the elasticsearch side, summary of the larger changes:
API changes: postings API became a "pull" rather than "push", collector API became per-segment, etc.
packaging changes: add lucene-backwards-codecs.jar as a dependency.
improvements to boolean filtering: especially ensuring it will not be slow for SparseBitSet.
use generic BitSet api in plumbing so that concrete bitset type is an implementation detail.
use generic BitDocIdSetFilter api for dedicated bitset cache, so there is type safety.
changes to support atomic commits
implement Accountable.getChildResources (detailed memory usage API) for fielddata, etc
change handling of IndexFormatTooOld/New, since they no longer extends CorruptIndexException
Closes #8347.
Squashed commit of the following:
commit d90d53f5f21b876efc1e09cbd6d63c538a16cd89
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Nov 5 21:35:28 2014 +0100
Make default codec/postings/docvalues format constants
commit cb66c22c71cd304a36e7371b199a8c279908ae37
Merge: d4e2f6d ad4ff43
Author: Robert Muir <rmuir@apache.org>
Date: Wed Nov 5 11:41:13 2014 -0500
Merge branch 'master' into enhancement/lucene_5_0_upgrade
commit d4e2f6dfe767a5128c9b9ae9e75036378de08f47
Merge: 4e5445c 4111d93
Author: Robert Muir <rmuir@apache.org>
Date: Wed Nov 5 06:26:32 2014 -0500
Merge branch 'master' into enhancement/lucene_5_0_upgrade
commit 4e5445c775f580730eb01360244e9330c0dc3958
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 16:19:19 2014 -0500
FixedBitSet -> BitSet
commit 9887ea73e8b857eeda7f851ef3722ef580c92acf
Merge: 1bf8894 fc84666
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 15:26:25 2014 -0500
Merge branch 'master' into enhancement/lucene_5_0_upgrade
commit 1bf8894430de3e566d0dc5623b0cc28b0d674ebb
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 15:22:51 2014 -0500
remove nocommit
commit a9c2a2259ff79c69bae7806b64e92d5f472c18c8
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 13:48:43 2014 -0500
turn jenkins red again
commit 067baaaa4d52fce772c81654dcdb5051ea79139f
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 13:18:21 2014 -0500
unzip from stream
commit 82b6fba33d362aca2313cc0ca495f28f5ebb9260
Merge: b2214bb 6523cd9
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 13:10:59 2014 -0500
Merge branch 'master' into enhancement/lucene_5_0_upgrade
commit b2214bb093ec2f759003c488c3c403c8931db914
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 13:09:53 2014 -0500
go back to my URL until we can figure out what is up with jenkins
commit e7d614172240175a51f580aeaefb6460d21cede9
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 10:52:54 2014 -0500
try this jenkins
commit 337a3c7704efa7c9809bf373152d711ee55f876c
Author: Simon Willnauer <simonw@apache.org>
Date: Tue Nov 4 16:17:49 2014 +0100
Rename temp-files under lock to prevent metadata reads while renaming
commit 77d5ba80d0a76efa549dd753b9f114b2f2d2d29c
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 10:07:11 2014 -0500
continue to treat too-old/too-new as corruption for now
commit 98d0fd2f4851bc50e505a94ca592a694d502c51c
Author: Robert Muir <rmuir@apache.org>
Date: Tue Nov 4 09:24:21 2014 -0500
fix last nocommit
commit 643fceed66c8caf22b97fc489d67b4a2a90a1a1c
Author: Simon Willnauer <simonw@apache.org>
Date: Tue Nov 4 14:46:17 2014 +0100
remove NoSuchDirectoryException
commit 2e43c4feba05cfaf451df70f946c0930cbcc4557
Merge: 93826e4 8163107
Author: Simon Willnauer <simonw@apache.org>
Date: Tue Nov 4 14:38:00 2014 +0100
Merge branch 'master' into enhancement/lucene_5_0_upgrade
commit 93826e4d56a6a97c2074669014af77ff519bde63
Merge: 7f10129 44e24d3
Author: Simon Willnauer <simonw@apache.org>
Date: Tue Nov 4 12:54:27 2014 +0100
Merge branch 'master' into enhancement/lucene_5_0_upgrade
Conflicts:
src/main/java/org/elasticsearch/index/store/DistributorDirectory.java
src/main/java/org/elasticsearch/index/store/Store.java
src/main/java/org/elasticsearch/indices/recovery/RecoveryStatus.java
src/test/java/org/elasticsearch/index/store/DistributorDirectoryTest.java
src/test/java/org/elasticsearch/index/store/StoreTest.java
src/test/java/org/elasticsearch/indices/recovery/RecoveryStatusTests.java
commit 7f10129364623620575c109df725cf54488b3abb
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Nov 4 11:32:24 2014 +0100
Fix TopHitsAggregator to not ignore the top-level/leaf collector split.
commit 042fadc8603b997bdfdc45ca44fec70dc86774a6
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Nov 4 11:31:20 2014 +0100
Remove MatchDocIdSet in favor of DocValuesDocIdSet.
commit 7d877581ff5db585a674c95ac391ac78a0282826
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Nov 4 11:10:08 2014 +0100
Make the and filter use the cost API.
Lucene 5 ensured that cost() can safely be used, and this will have the benefit
that the order in which filters are specified is not important anymore (only
for slow random-access filters in practice).
commit 78f1718aa2cd82184db7c3a8393e6215f43eb4a8
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 23:55:17 2014 -0500
fix previous eclipse import braindamage
commit 186c40e9258ce32f22a9a714ab442a310b6376e0
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 22:32:34 2014 -0500
allow child queries to exhaust iterators again
commit b0b1271305e1b6d0c4c4da51a3c54df1aa5c0605
Author: Ryan Ernst <ryan@iernst.net>
Date: Mon Nov 3 14:50:44 2014 -0800
Fix nocommit for mapping output. index_options will not be printed if
the field is not indexed.
commit ba223eb85e399c9620a347a983e29bf703953e7a
Author: Ryan Ernst <ryan@iernst.net>
Date: Mon Nov 3 14:07:26 2014 -0800
Remove no commit for chinese analyzer provider. We should have a
separate issue to address not using this provider on new indexes.
commit ca554b03c4471797682b2fb724f25205cf040c4a
Author: Ryan Ernst <ryan@iernst.net>
Date: Mon Nov 3 13:41:59 2014 -0800
Fix stop tests
commit de67c4653ec47dee9c671390536110749d2bb05f
Author: Ryan Ernst <ryan@iernst.net>
Date: Mon Nov 3 12:51:17 2014 -0800
Remove analysis nocommits, switching over to Lucene43*Filters for
backcompat
commit 50cae9bec72c25c33a1ab8a8931bccb3355171e2
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 15:32:25 2014 -0500
add ram accounting and TODO lazy-loading (its no worse than master, can be a followup improvement) for suggesters
commit 7a7f0122f138684b312d0f0b03dc2a9c16c15f9c
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 15:11:26 2014 -0500
bump lucene version
commit cd0cae5c35e7a9e049f49ae45431f658fb86676b
Merge: 446bc09 3c72073
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 14:49:05 2014 -0500
Merge branch 'master' into enhancement/lucene_5_0_upgrade
commit 446bc09b4e8bf4602d3c252b53ddaa0da65cce2f
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 14:46:30 2014 -0500
remove hack
commit a19d85a968d82e6d00292b49630ef6ff2dbf2f32
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 12:53:11 2014 -0500
dont create exceptions with circular references on corruption (will open a PR for this)
commit 0beefb9e821d97c37e90ec556d81ac7b00369b8a
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 11:47:14 2014 -0500
temporarily add craptastic detector for this horrible bug
commit e9f2d298bff75f3d1591f8622441e459c3ce7ac3
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 10:56:01 2014 -0500
add nocommit
commit e97f1d50a91a7129650b8effc7a9ecf74ca0569a
Merge: c57a3c8 f1f50ac
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 10:12:12 2014 -0500
Merge branch 'master' into enhancement/lucene_5_0_upgrade
commit c57a3c8341ed61dca62eaf77fad6b8b48aeb6940
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 10:11:46 2014 -0500
fix nocommit
commit dd0e77e4ec07c7011ab5f6b60b2ead33dc2333d2
Author: Robert Muir <rmuir@apache.org>
Date: Mon Nov 3 09:54:09 2014 -0500
nocommit -> TODO, this is in much more places in the codebase, bigger issue
commit 3cc3bf56d72d642059f8fe220d6f2fed608363e9
Author: Ryan Ernst <ryan@iernst.net>
Date: Sat Nov 1 23:59:17 2014 -0700
Remove nocommit and awaitsfix for edge ngram filter test.
commit 89f115245155511c0fbc0d5ee62e63141c3700c1
Author: Ryan Ernst <ryan@iernst.net>
Date: Sat Nov 1 23:57:44 2014 -0700
Fix EdgeNGramTokenFilter logic for version <= 4.3, and fixed instanceof
checks in corresponding tests to correctly check for reverse filter when
applicable.
commit 112df869cd199e36aab0e1a7a288bb1fdb2ebf1c
Author: Robert Muir <rmuir@apache.org>
Date: Sun Nov 2 00:08:30 2014 -0400
execute geo disjoint query/filter as intersects
commit e5061273cc685f1252e9a3a9ae4877ec9bce7752
Author: Robert Muir <rmuir@apache.org>
Date: Sat Nov 1 22:58:59 2014 -0400
remove chinese analyzer from docs
commit ea1af11b8978fcc551f198e24fe21d52806993ef
Author: Robert Muir <rmuir@apache.org>
Date: Sat Nov 1 22:29:00 2014 -0400
fix ram accounting bug
commit 53c0a42c6aa81aa6bf81d3aa77b95efd513e0f81
Merge: e3bcd3c 6011a18
Author: Robert Muir <rmuir@apache.org>
Date: Sat Nov 1 22:16:29 2014 -0400
Merge branch 'master' into enhancement/lucene_5_0_upgrade
commit e3bcd3cc07a4957e12c7b3affc462c31290a9186
Author: Robert Muir <rmuir@apache.org>
Date: Sat Nov 1 22:15:01 2014 -0400
fix url-email back compat (thanks ryan)
commit 91d6b096a96c357755abee167098607223be1aad
Author: Robert Muir <rmuir@apache.org>
Date: Sat Nov 1 22:11:26 2014 -0400
bump lucene version
commit d2bb9568df72b37ec7050d25940160b8517394bc
Author: Robert Muir <rmuir@apache.org>
Date: Sat Nov 1 20:33:07 2014 -0400
remove nocommit
commit 1d049c471e19e5c457262c7399c5bad9e023b2e3
Author: Robert Muir <rmuir@apache.org>
Date: Sat Nov 1 20:28:58 2014 -0400
fix eclipse to group org/com imports together: without this, its madness
commit 09d8c1585ee99b6e63be032732c04ef6fed84ed2
Author: Robert Muir <rmuir@apache.org>
Date: Sat Nov 1 14:27:41 2014 -0400
remove nocommit, if you dont liek it, print assembly and tell me how it can be better
commit 8a6a294313fdf33b50c7126ec20c07867ecd637c
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 31 20:01:55 2014 +0100
Remove deprecated usage of DocIdSets.newDocIDSet.
commit 601bee60543610558403298124a84b1b3bbd1045
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 31 14:13:18 2014 -0400
maybe one of these zillions of annotations will stop thread leaks
commit 9d3f69abc7267c5e455aefa26db95cb554b02d62
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 31 14:05:39 2014 -0400
fix some analysis nocommits
commit 312e3a29c77214b8142d21c33a6b2c2b151acf9a
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 31 18:28:45 2014 +0100
Remove XConstantScoreQuery/XFilteredQuery/ApplyAcceptedDocsFilter.
commit 5a0cb9f8e167215df7f1b1fad11eec6e6c74940f
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 31 17:06:45 2014 +0100
Fix misleading documentation of DocIdSets.toCacheable.
commit 8b4ef2b5b476fff4c79c0c2a0e4769ead26cf82b
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 31 17:05:59 2014 +0100
Fix CustomRandomAccessFilterStrategy to override the right method.
commit d7a9a407a615987cfffc651f724fbd8795c9c671
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 31 16:21:35 2014 +0100
Better handle the special case when there is a single SHOULD clause.
commit 648ad389f07e92dfc451f345549c9841ba5e4c9a
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 31 15:53:38 2014 +0100
Cut over XBooleanFilter to BitDocIdSet.Builder.
The idea is similar to what happened to Lucene's BooleanFilter.
Yet XBooleanFilter is a bit more sophisticated and I had to slightly
change the way it is implemented in order to make it work. The main difference
with before is that slow filters are now applied lazily, so eg. if you have 3
MUST clauses, two with a fast iterator and the third with a slow iterator, the
previous implementation used to apply the fast iterators first and then only
check the slow filter for bits which were set in the bit set. Now we are
computing a bit set based on the fast must clauses and then basically returning
a BitsFilteredDocIdSet.wrap(bitset, slowClause).
Other than that, BooleanFilter still uses the bitset optimizations when or-ing
and and-ind filters.
Another improvement is that BooleanFilter is now aware of the cost API.
commit b2dad312b4bc9f931dc3a25415dd81c0d9deee08
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 31 10:18:53 2014 -0400
clear nocommit
commit 4851d2091e744294336dfade33906c75fbe695cd
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 15:15:16 2014 +0100
cut over to RoaringDocIdSet
commit ca6aec24a901073e65ce4dd6b70964fd3612409e
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 14:57:30 2014 +0100
make nocommit more explicit
commit d0742ee2cb7a6c48b0bbb31580b7fbcebdb6ec40
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 31 09:55:24 2014 -0400
fix standardtokenizer nocommit
commit 7d6faccafff22a86af62af0384838391d46695ca
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 14:54:08 2014 +0100
fix compilation
commit a038a405c1ff6458ad294e6b5bc469e622f699d0
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 14:53:43 2014 +0100
fix compilation
commit 30c9e307b1f5d80e2deca3392c0298682241207f
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 14:52:35 2014 +0100
fix compilation
commit e5139bc5a0a9abd2bdc6ba0dfbcb7e3c2e7b8481
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 31 09:52:16 2014 -0400
clear nocommit here
commit 85dd2cedf7a7994bed871ac421cfda06aaf5c0a5
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 14:46:17 2014 +0100
fix CompletionPostingsFormatTest
commit c0f3781f616c9b0ee3b5c4d0998810f595868649
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 31 09:38:00 2014 -0400
add tests for these analyzers
commit 51f9999b4ad079c283ae762c862fd0e22d00445f
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 14:10:26 2014 +0100
remove nocommit - this is not an issue
commit fd1388fa03e622b0738601c8aeb2dbf7949a6dd2
Author: Martijn van Groningen <martijn.v.groningen@gmail.com>
Date: Fri Oct 31 14:07:01 2014 +0100
Remove redundant null check
commit 3d6dd51b0927337ba941a235446b22e8cd500dc3
Author: Martijn van Groningen <martijn.v.groningen@gmail.com>
Date: Fri Oct 31 14:01:37 2014 +0100
Removed the work around to prevent p/c error when invoking #iterator() twice, because the custom query filter wrapper now doesn't transform the result to a cache doc id set any more.
I think the transforming to a cachable doc id set in CustomQueryWrappingFilter isn't needed at all, because we use the DocIdSet only once and because of that is just slowed things down.
commit 821832a537e00cd1216064b379df3e01d2911d3a
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 13:54:33 2014 +0100
one more nocommit
commit 77eb9ea4c4ea50afb2680c29682ddcb3851a9d4f
Author: Martijn van Groningen <martijn.v.groningen@gmail.com>
Date: Fri Oct 31 13:52:29 2014 +0100
Remove cast
commit a400573c034ed602221f801b20a58a9186a06eae
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 13:49:24 2014 +0100
fix stop filter
commit 51746087cf8ec34c4d20aa05ba8dbff7b3b43eec
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 13:21:36 2014 +0100
fix changed semantics of FBS.nextSetBit to check for NO_MORE_DOCS
commit 8d0a4e2511310f1293860823fe3ba80ac771bbe3
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 31 08:13:44 2014 -0400
do the bogus cast differently
commit 46a5cc5732dea096c0c80ae5ce42911c9c51e44e
Author: Simon Willnauer <simonw@apache.org>
Date: Fri Oct 31 13:00:16 2014 +0100
I hate it but P/C now passes
commit 580c0c2f82bbeacf217e594f22312b11d1bdb839
Merge: a9d3c00 1645434
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 31 06:54:31 2014 -0400
fix nocommit/classcast
commit a9d3c004d62fe04989f49a897e6ff84973c06eb9
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 31 08:49:31 2014 +0100
Update TODO.
commit aa75af0b407792aeef32017f03a6f442ed970baa
Author: Robert Muir <rmuir@apache.org>
Date: Thu Oct 30 19:18:25 2014 -0400
clear obselete nocommits from lucene bump
commit d438534cf41fcbe2d88070e2f27c994625e082c2
Author: Robert Muir <rmuir@apache.org>
Date: Thu Oct 30 18:53:20 2014 -0400
throw classcastexception when ES abuses regular filtercache for nested docs
commit 2c751f3a8feda43ec127c34769b069de21f3d16f
Author: Robert Muir <rmuir@apache.org>
Date: Thu Oct 30 18:31:34 2014 -0400
bump lucene revision, fix tests
commit d6ef7f6304ae262bf6228a7d661b2a452df332be
Author: Simon Willnauer <simonw@apache.org>
Date: Thu Oct 30 22:37:58 2014 +0100
fix merge problems
commit de9d361f88a9ce6bb3fba85285de41f223c95767
Merge: 41f6aab f6b37a3
Author: Simon Willnauer <simonw@apache.org>
Date: Thu Oct 30 22:28:59 2014 +0100
Merge branch 'master' into enhancement/lucene_5_0_upgrade
Conflicts:
pom.xml
src/main/java/org/elasticsearch/Version.java
src/main/java/org/elasticsearch/gateway/local/state/meta/MetaDataStateFormat.java
commit 41f6aab388aa80c40b08a2facab2617576203a0d
Author: Simon Willnauer <simonw@apache.org>
Date: Thu Oct 30 17:48:46 2014 +0100
fix potiential NPE
commit c4428b12e1ae838b91e847df8b4a8be7f49e10f4
Author: Simon Willnauer <simonw@apache.org>
Date: Thu Oct 30 17:38:46 2014 +0100
don't advance iterator in a match(doc) method
commit 28ab948e99e3ea4497c9b1e468384806ba7e1790
Author: Simon Willnauer <simonw@apache.org>
Date: Thu Oct 30 17:34:58 2014 +0100
don't advance iterator in a match(doc) method
commit eb0f33f6634fadfcf4b2bf7327400e568f0427bb
Author: Simon Willnauer <simonw@apache.org>
Date: Thu Oct 30 16:55:54 2014 +0100
fix GeoUtilsTest
commit 7f711fe3eaf73b6c2268cf42d5a41132a61ad831
Author: Simon Willnauer <simonw@apache.org>
Date: Thu Oct 30 16:43:16 2014 +0100
Use a dedicated default index option if field type is not indexed by default
commit 78e3f37ab779e3e1b25b45a742cc86ab5f975149
Author: Robert Muir <rmuir@apache.org>
Date: Thu Oct 30 10:56:14 2014 -0400
disable this test with AwaitsFix to reduce noise
commit 9a590f563c8e03a99ecf0505c92d12d7ab20d11d
Author: Simon Willnauer <simonw@apache.org>
Date: Thu Oct 30 09:38:49 2014 +0100
fix lucene version
commit abe3ca1d8bb6b5101b545198f59aec44bacfa741
Author: Simon Willnauer <simonw@apache.org>
Date: Thu Oct 30 09:35:05 2014 +0100
fix AnalyzingCompletionLookupProvider to wrok with new codec API
commit 464293b245852d60bde050c6d3feb5907dcfbf5f
Author: Robert Muir <rmuir@apache.org>
Date: Thu Oct 30 00:26:00 2014 -0400
don't try to write stuff to tests class directory
commit 031cc6c19f4fe4423a034b515f77e5a0e282a124
Author: Robert Muir <rmuir@apache.org>
Date: Thu Oct 30 00:12:36 2014 -0400
AwaitsFix these known issues to reduce noise
commit 4600d51891e35847f2d344247d6f915a0605c0d1
Author: Robert Muir <rmuir@apache.org>
Date: Thu Oct 30 00:06:53 2014 -0400
openbitset lives on
commit 8492bae056249e2555d24acd55f1046b66a667c4
Author: Robert Muir <rmuir@apache.org>
Date: Wed Oct 29 23:42:54 2014 -0400
fixes for filter tests
commit 31f24ce4efeda31f97eafdb122346c7047a53bf2
Author: Robert Muir <rmuir@apache.org>
Date: Wed Oct 29 23:12:38 2014 -0400
don't use fieldcache
commit 8480789942fdff14a6d2b2cd8134502fe62f20c8
Author: Robert Muir <rmuir@apache.org>
Date: Wed Oct 29 23:04:29 2014 -0400
ancient index no longer supported
commit 02e78dc7ebdd827533009f542582e8db44309c57
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 23:37:02 2014 +0100
fix more tests
commit ff746c6df23c50b3f3ec24922413b962c8983080
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 23:08:19 2014 +0100
fix all mapper
commit e4fb84b517107b25cb064c66f83c9aa814a311b2
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 22:55:54 2014 +0100
fix distributor tests and cut over to FileStore API
commit 20c850e2cfe3210cd1fb9e232afed8d4ac045857
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 22:42:18 2014 +0100
use DOCS_ONLY if index=true and current options == null
commit 44169c108418413cfe51f5ce23ab82047463e4c2
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 22:33:36 2014 +0100
Fix index=yes|no settings in mappers
commit a3c5f77987461a18121156ed345d42ded301c566
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 21:51:41 2014 +0100
fix several field mappers conversion from setIndexed to indexOptions
commit df84d736908e88a031d710f98e222be68ae96af1
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 21:33:35 2014 +0100
fix SourceFieldMapper to be not indexed
commit b2bf01d12a8271a31fb2df601162d0e89924c8f5
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 21:23:08 2014 +0100
Cut over to .liv files in store and corruption tests
commit 619004df436f9ef05d24bef1b6a7f084c6b0ad75
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 17:05:52 2014 +0100
fix more tests
commit b7ed653a8b464de446e00456bce0a89e47627c38
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 16:19:08 2014 +0100
[STORE] Add dedicated method to write temporary files
Recovery writes temporary files which might not end up in the
right distributor directories today. This commit adds a dedicated
API that allows specifying the target file name in order to create the
tempoary file in the correct directory.
commit 7d574659f6ae04adc2b857146ad0d8d56ca66f12
Author: Robert Muir <rmuir@apache.org>
Date: Wed Oct 29 10:28:49 2014 -0400
add some leniency to temporary bogus method
commit f97022ea7c2259f7a5cf97d924c59ed75ab65b32
Author: Robert Muir <rmuir@apache.org>
Date: Wed Oct 29 10:24:17 2014 -0400
fix MultiCollector bug
commit b760533128c2b4eb10ad76e9689ef714293dd819
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 14:56:08 2014 +0100
CheckIndex is now closeable we need to close it
commit 9dae9fb6d63546a6c2427be2a2d5c8358f5b1934
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 14:45:11 2014 +0100
s/Lucene51/Lucene50
commit 7aea9b86856a8c1b06a08e7c312ede1168af1287
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 14:42:30 2014 +0100
fix BloomFilterPostingsFormat
commit 16fea6fe842e88665d59cc091e8224e8dc6ce08c
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 14:41:16 2014 +0100
fix some codec format issues
commit 3d77aa97dd2c4012b63befef3f2ba2525965e8a6
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 14:30:43 2014 +0100
fix CodecTests
commit 6ef823b1fde25657438ace1aabd9d552d6ae215e
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 14:26:47 2014 +0100
make it compile
commit 9991eee1fe99435118d4dd42b297ffc83fce5ec5
Author: Robert Muir <rmuir@apache.org>
Date: Wed Oct 29 09:12:43 2014 -0400
add an ugly hack for TopHitsAggregator for now
commit 03e768a01fcae6b1f4cb50bcceec7d42977ac3e6
Author: Simon Willnauer <simonw@apache.org>
Date: Wed Oct 29 14:01:02 2014 +0100
cut over ES090PostingsFormat
commit 463d281faadb794fdde3b469326bdaada25af048
Merge: 0f8740a 8eac79c
Author: Robert Muir <rmuir@apache.org>
Date: Wed Oct 29 08:30:36 2014 -0400
Merge branch 'master' into enhancement/lucene_5_0_upgrade
commit 0f8740a782455a63524a5a82169f6bbbfc613518
Author: Robert Muir <rmuir@apache.org>
Date: Wed Oct 29 01:00:15 2014 -0400
fix/hack remaining filter and analysis issues
commit df534488569da13b31d66e581456dfd4b55156b9
Author: Robert Muir <rmuir@apache.org>
Date: Tue Oct 28 23:11:47 2014 -0400
fix ngrams / openbitset usage
commit 11f5dc3b9887f4da80a0fa1818e1350b30599329
Author: Robert Muir <rmuir@apache.org>
Date: Tue Oct 28 22:42:44 2014 -0400
hack over sort comparators
commit 4ebdc754350f512596f6a02770d223e9f5f7975a
Author: Robert Muir <rmuir@apache.org>
Date: Tue Oct 28 21:27:07 2014 -0400
compiler errors < 100
commit 2d60c9e29de48ccb0347dd87f7201f47b67b83a0
Author: Robert Muir <rmuir@apache.org>
Date: Tue Oct 28 03:13:08 2014 -0400
clear some nocommits around ram usage
commit aaf47fe6c0aabcfb2581dd456fc50edf871da758
Author: Robert Muir <rmuir@apache.org>
Date: Mon Oct 27 12:27:34 2014 -0400
migrate fieldinfo handling
commit ef6ed6d15d8def71cd880d97249678136cd29fe3
Author: Robert Muir <rmuir@apache.org>
Date: Mon Oct 27 12:07:13 2014 -0400
more simple fixes
commit f475e1048ae697dd9da5bd9da445102b0b7bc5b3
Author: Robert Muir <rmuir@apache.org>
Date: Mon Oct 27 11:58:21 2014 -0400
more fielddata ram accounting fixes
commit 16b4239eaa9b4262df258257df4f31d39f28a3a2
Author: Simon Willnauer <simonw@apache.org>
Date: Mon Oct 27 16:47:32 2014 +0100
add missing file
commit 5b542fa2a6da81e36a0c35b8e891a1d8bc58f663
Author: Simon Willnauer <simonw@apache.org>
Date: Mon Oct 27 16:43:29 2014 +0100
cut over completion posting formats - still some nocommits
commit ecdea49404c4ec4e1b78fb54575825f21b4e096e
Author: Robert Muir <rmuir@apache.org>
Date: Mon Oct 27 11:21:09 2014 -0400
fielddata accountable fixes
commit d43da265718917e20c8264abd43342069198fe9c
Author: Simon Willnauer <simonw@apache.org>
Date: Mon Oct 27 16:19:53 2014 +0100
cut over BloomFilterPostings to new API
commit 29b192ba621c14820175775d01242162b88bd364
Author: Robert Muir <rmuir@apache.org>
Date: Mon Oct 27 10:22:51 2014 -0400
fix more analyzers
commit 74b4a0c5283e323a7d02490df469497c722780d2
Author: Robert Muir <rmuir@apache.org>
Date: Mon Oct 27 09:54:25 2014 -0400
fix tests
commit 554084ccb4779dd6b1c65fa7212ad1f64f3a6968
Author: Simon Willnauer <simonw@apache.org>
Date: Mon Oct 27 14:51:48 2014 +0100
maintain supressed exceptions on CorruptIndexException
commit cf882d9112c5e8ef1e9f2b0f800f7aa59001a4f2
Author: Simon Willnauer <simonw@apache.org>
Date: Mon Oct 27 14:47:17 2014 +0100
commitOnClose=false
commit ebb2a9189ab2f459b7c6c9985be610fd90dfe410
Author: Simon Willnauer <simonw@apache.org>
Date: Mon Oct 27 14:46:06 2014 +0100
cut over indexwriter closeing in InternalEngine
commit cd21b3d4706f0b562bd37792d077d60832aff65f
Author: Simon Willnauer <simonw@apache.org>
Date: Mon Oct 27 14:38:10 2014 +0100
fix constant
commit f93f900c4a1c90af3a21a4af5735a7536423fe28
Author: Robert Muir <rmuir@apache.org>
Date: Mon Oct 27 09:50:49 2014 -0400
fix test
commit a9a752940b1ab4699a6a08ba8b34afca82b843fe
Author: Martijn van Groningen <martijn.v.groningen@gmail.com>
Date: Mon Oct 27 09:26:18 2014 +0100
Be explicit about the index options
commit d9ee815babd030fa2ceaec9f467c105ee755bf6b
Author: Simon Willnauer <simonw@apache.org>
Date: Sun Oct 26 20:03:44 2014 +0100
cut over store and directory
commit b3f5c8e39039dd8f5caac0c4dd1fc3b1116e64ca
Author: Robert Muir <rmuir@apache.org>
Date: Sun Oct 26 13:08:39 2014 -0400
more test fixes
commit 8842f2684e3606aae0860c27f7a4c53e273d47fb
Author: Robert Muir <rmuir@apache.org>
Date: Sun Oct 26 12:14:52 2014 -0400
tests manual labor
commit c43de5aec337919a3fdc3638406dff17fc80bc98
Author: Robert Muir <rmuir@apache.org>
Date: Sun Oct 26 11:04:13 2014 -0400
BytesRef -> BytesRefBuilder
commit 020c0d087a2f37566a1db390b0e044ebab030138
Author: Martijn van Groningen <martijn.v.groningen@gmail.com>
Date: Sun Oct 26 15:53:37 2014 +0100
Moved over to BitSetFilter
commit 48dd1b909e6c52cef733961c9ecebfe4f67109fe
Author: Martijn van Groningen <martijn.v.groningen@gmail.com>
Date: Sun Oct 26 15:53:11 2014 +0100
Left over Collector api change in ScanContext
commit 6ec248ef63f262bcda400181b838fd9244752625
Author: Martijn van Groningen <martijn.v.groningen@gmail.com>
Date: Sun Oct 26 15:47:40 2014 +0100
Moved indexed() over to indexOptions != null or indexOptions == null
commit 9937aebfd8546ae4bb652cd976b3b43ac5ab7a63
Author: Martijn van Groningen <martijn.v.groningen@gmail.com>
Date: Sun Oct 26 13:26:31 2014 +0100
Fixed many compile errors. Mainly around the breaking Collector api change in 5.0.
commit fec32c4abc0e3309cf34260c8816305a6f820c9e
Author: Robert Muir <rmuir@apache.org>
Date: Sat Oct 25 11:22:17 2014 -0400
more easy fixes
commit dab22531d801800d17a65dc7c9464148ce8ebffd
Author: Robert Muir <rmuir@apache.org>
Date: Sat Oct 25 09:33:41 2014 -0400
more progress
commit 414767e9a955010076b0497cc4f6d0c1850b48d3
Author: Robert Muir <rmuir@apache.org>
Date: Sat Oct 25 06:33:17 2014 -0400
more progress
commit ad9d969fddf139a8830254d3eb36a908ba87cc12
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 24 14:28:01 2014 -0400
current state of fun
commit 464475eecb0be15d7d084135ed16051f76a7e521
Author: Robert Muir <rmuir@apache.org>
Date: Fri Oct 24 11:42:41 2014 -0400
bump to 5.0 snapshot
2014-11-05 15:48:51 -05:00
|
|
|
public BitsetFilterCache bitsetFilterCache() {
|
2015-11-04 11:42:18 +01:00
|
|
|
return indexService.cache().bitsetFilterCache();
|
2014-07-25 16:17:48 +02:00
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2017-07-26 09:16:52 +02:00
|
|
|
public <IFD extends IndexFieldData<?>> IFD getForField(MappedFieldType fieldType) {
|
|
|
|
return queryShardContext.getForField(fieldType);
|
2013-08-26 16:28:27 +02:00
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2016-06-27 18:41:18 -04:00
|
|
|
public TimeValue timeout() {
|
|
|
|
return timeout;
|
2013-08-26 16:28:27 +02:00
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2016-06-27 18:41:18 -04:00
|
|
|
public void timeout(TimeValue timeout) {
|
|
|
|
this.timeout = timeout;
|
2013-08-26 16:28:27 +02:00
|
|
|
}
|
|
|
|
|
2014-07-18 10:53:20 -04:00
|
|
|
@Override
|
|
|
|
public int terminateAfter() {
|
|
|
|
return terminateAfter;
|
|
|
|
}
|
|
|
|
|
|
|
|
@Override
|
|
|
|
public void terminateAfter(int terminateAfter) {
|
|
|
|
this.terminateAfter = terminateAfter;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public SearchContext minimumScore(float minimumScore) {
|
|
|
|
this.minimumScore = minimumScore;
|
|
|
|
return this;
|
|
|
|
}
|
|
|
|
|
2015-10-26 18:37:35 +01:00
|
|
|
@Override
|
|
|
|
public Float minimumScore() {
|
|
|
|
return this.minimumScore;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2016-04-25 17:45:33 +02:00
|
|
|
public SearchContext sort(SortAndFormats sort) {
|
2013-08-26 16:28:27 +02:00
|
|
|
this.sort = sort;
|
|
|
|
return this;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2016-04-25 17:45:33 +02:00
|
|
|
public SortAndFormats sort() {
|
2013-08-26 16:28:27 +02:00
|
|
|
return this.sort;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public SearchContext trackScores(boolean trackScores) {
|
|
|
|
this.trackScores = trackScores;
|
|
|
|
return this;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public boolean trackScores() {
|
|
|
|
return this.trackScores;
|
|
|
|
}
|
|
|
|
|
2017-06-08 12:10:46 +02:00
|
|
|
@Override
|
2019-01-04 20:36:49 +01:00
|
|
|
public SearchContext trackTotalHitsUpTo(int trackTotalHitsUpTo) {
|
|
|
|
this.trackTotalHitsUpTo = trackTotalHitsUpTo;
|
2017-06-08 12:10:46 +02:00
|
|
|
return this;
|
|
|
|
}
|
|
|
|
|
|
|
|
@Override
|
2019-01-04 20:36:49 +01:00
|
|
|
public int trackTotalHitsUpTo() {
|
|
|
|
return trackTotalHitsUpTo;
|
2017-06-08 12:10:46 +02:00
|
|
|
}
|
|
|
|
|
2016-01-12 17:40:34 +01:00
|
|
|
@Override
|
|
|
|
public SearchContext searchAfter(FieldDoc searchAfter) {
|
|
|
|
this.searchAfter = searchAfter;
|
|
|
|
return this;
|
|
|
|
}
|
|
|
|
|
2016-10-25 12:27:33 -10:00
|
|
|
@Override
|
|
|
|
public boolean lowLevelCancellation() {
|
|
|
|
return lowLevelCancellation;
|
|
|
|
}
|
|
|
|
|
2016-01-12 17:40:34 +01:00
|
|
|
@Override
|
|
|
|
public FieldDoc searchAfter() {
|
|
|
|
return searchAfter;
|
|
|
|
}
|
|
|
|
|
2017-01-23 16:33:51 +01:00
|
|
|
@Override
|
|
|
|
public SearchContext collapse(CollapseContext collapse) {
|
|
|
|
this.collapse = collapse;
|
|
|
|
return this;
|
|
|
|
}
|
|
|
|
|
|
|
|
@Override
|
|
|
|
public CollapseContext collapse() {
|
|
|
|
return collapse;
|
|
|
|
}
|
|
|
|
|
2016-06-08 15:28:06 +02:00
|
|
|
public SearchContext sliceBuilder(SliceBuilder sliceBuilder) {
|
|
|
|
this.sliceBuilder = sliceBuilder;
|
2016-05-10 12:10:55 +02:00
|
|
|
return this;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2015-05-05 08:27:52 +02:00
|
|
|
public SearchContext parsedPostFilter(ParsedQuery postFilter) {
|
2013-12-16 12:11:06 +01:00
|
|
|
this.postFilter = postFilter;
|
2013-08-26 16:28:27 +02:00
|
|
|
return this;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2015-05-05 08:27:52 +02:00
|
|
|
public ParsedQuery parsedPostFilter() {
|
2013-12-16 12:11:06 +01:00
|
|
|
return this.postFilter;
|
2013-08-26 16:28:27 +02:00
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2015-05-05 08:27:52 +02:00
|
|
|
public Query aliasFilter() {
|
2013-08-26 16:28:27 +02:00
|
|
|
return aliasFilter;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public SearchContext parsedQuery(ParsedQuery query) {
|
|
|
|
this.originalQuery = query;
|
|
|
|
this.query = query.query();
|
|
|
|
return this;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public ParsedQuery parsedQuery() {
|
|
|
|
return this.originalQuery;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**
|
2015-08-14 13:29:33 +02:00
|
|
|
* The query to execute, in its rewritten form.
|
2013-08-26 16:28:27 +02:00
|
|
|
*/
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public Query query() {
|
|
|
|
return this.query;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public int from() {
|
|
|
|
return from;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public SearchContext from(int from) {
|
|
|
|
this.from = from;
|
|
|
|
return this;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public int size() {
|
|
|
|
return size;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public SearchContext size(int size) {
|
|
|
|
this.size = size;
|
|
|
|
return this;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2016-08-17 15:59:38 +02:00
|
|
|
public boolean hasStoredFields() {
|
|
|
|
return storedFields != null && storedFields.fieldNames() != null;
|
2013-08-26 16:28:27 +02:00
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2016-08-17 15:59:38 +02:00
|
|
|
public boolean hasStoredFieldsContext() {
|
|
|
|
return storedFields != null;
|
|
|
|
}
|
|
|
|
|
|
|
|
@Override
|
|
|
|
public StoredFieldsContext storedFieldsContext() {
|
|
|
|
return storedFields;
|
|
|
|
}
|
|
|
|
|
|
|
|
@Override
|
|
|
|
public SearchContext storedFieldsContext(StoredFieldsContext storedFieldsContext) {
|
|
|
|
this.storedFields = storedFieldsContext;
|
|
|
|
return this;
|
2013-08-26 16:28:27 +02:00
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2016-08-17 15:59:38 +02:00
|
|
|
public boolean storedFieldsRequested() {
|
|
|
|
return storedFields == null || storedFields.fetchFields();
|
2013-08-26 16:28:27 +02:00
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public boolean explain() {
|
|
|
|
return explain;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public void explain(boolean explain) {
|
|
|
|
this.explain = explain;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
@Nullable
|
|
|
|
public List<String> groupStats() {
|
|
|
|
return this.groupStats;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public void groupStats(List<String> groupStats) {
|
|
|
|
this.groupStats = groupStats;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public boolean version() {
|
|
|
|
return version;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public void version(boolean version) {
|
|
|
|
this.version = version;
|
|
|
|
}
|
|
|
|
|
2019-01-23 09:01:58 +01:00
|
|
|
@Override
|
|
|
|
public boolean seqNoAndPrimaryTerm() {
|
|
|
|
return seqAndPrimaryTerm;
|
|
|
|
}
|
|
|
|
|
|
|
|
@Override
|
|
|
|
public void seqNoAndPrimaryTerm(boolean seqNoAndPrimaryTerm) {
|
|
|
|
this.seqAndPrimaryTerm = seqNoAndPrimaryTerm;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public int[] docIdsToLoad() {
|
|
|
|
return docIdsToLoad;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public int docIdsToLoadFrom() {
|
|
|
|
return docsIdsToLoadFrom;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public int docIdsToLoadSize() {
|
|
|
|
return docsIdsToLoadSize;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public SearchContext docIdsToLoad(int[] docIdsToLoad, int docsIdsToLoadFrom, int docsIdsToLoadSize) {
|
|
|
|
this.docIdsToLoad = docIdsToLoad;
|
|
|
|
this.docsIdsToLoadFrom = docsIdsToLoadFrom;
|
|
|
|
this.docsIdsToLoadSize = docsIdsToLoadSize;
|
|
|
|
return this;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public void accessed(long accessTime) {
|
|
|
|
this.lastAccessTime = accessTime;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public long lastAccessTime() {
|
|
|
|
return this.lastAccessTime;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public long keepAlive() {
|
|
|
|
return this.keepAlive;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public void keepAlive(long keepAlive) {
|
|
|
|
this.keepAlive = keepAlive;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public DfsSearchResult dfsResult() {
|
|
|
|
return dfsResult;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public QuerySearchResult queryResult() {
|
|
|
|
return queryResult;
|
|
|
|
}
|
|
|
|
|
2015-12-16 10:22:54 +00:00
|
|
|
@Override
|
|
|
|
public FetchPhase fetchPhase() {
|
|
|
|
return fetchPhase;
|
|
|
|
}
|
|
|
|
|
2015-02-23 17:07:46 -05:00
|
|
|
@Override
|
2013-08-26 16:28:27 +02:00
|
|
|
public FetchSearchResult fetchResult() {
|
|
|
|
return fetchResult;
|
|
|
|
}
|
|
|
|
|
2015-01-29 13:22:46 -08:00
|
|
|
@Override
|
2020-06-17 10:36:31 -07:00
|
|
|
public MappedFieldType fieldType(String name) {
|
2020-02-07 10:24:01 -08:00
|
|
|
return mapperService().fieldType(name);
|
2015-01-29 13:22:46 -08:00
|
|
|
}
|
|
|
|
|
|
|
|
@Override
|
2015-06-15 15:36:27 -07:00
|
|
|
public ObjectMapper getObjectMapper(String name) {
|
2015-12-16 19:41:34 +01:00
|
|
|
return mapperService().getObjectMapper(name);
|
2013-08-26 16:28:27 +02:00
|
|
|
}
|
2014-01-31 01:46:02 +01:00
|
|
|
|
2014-09-22 22:15:59 +02:00
|
|
|
@Override
|
2019-02-21 12:42:54 +02:00
|
|
|
public long getRelativeTimeInMillis() {
|
|
|
|
return relativeTimeSupplier.getAsLong();
|
2014-09-22 22:15:59 +02:00
|
|
|
}
|
2014-10-13 09:41:55 +02:00
|
|
|
|
2015-08-14 13:29:33 +02:00
|
|
|
@Override
|
|
|
|
public Map<Class<?>, Collector> queryCollectors() {
|
|
|
|
return queryCollectors;
|
|
|
|
}
|
2015-09-25 17:12:57 +02:00
|
|
|
|
2016-02-08 12:09:15 +01:00
|
|
|
@Override
|
|
|
|
public QueryShardContext getQueryShardContext() {
|
|
|
|
return queryShardContext;
|
|
|
|
}
|
|
|
|
|
2015-09-25 17:12:57 +02:00
|
|
|
@Override
|
Add ability to profile query and collectors
Provides a new flag which can be enabled on a per-request basis.
When `"profile": true` is set, the search request will execute in a
mode that collects detailed timing information for query components.
```
GET /test/test/_search
{
"profile": true,
"query": {
"match": {
"foo": "bar"
}
}
}
```
Closes #14889
Squashed commit of the following:
commit a92db5723d2c61b8449bd163d2f006d12f9889ad
Merge: 117dd99 3f87b08
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Dec 17 09:44:10 2015 -0500
Merge remote-tracking branch 'upstream/master' into query_profiler
commit 117dd9992e8014b70203c6110925643769d80e62
Merge: 9b29d68 82a64fd
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Dec 15 13:27:18 2015 -0500
Merge remote-tracking branch 'upstream/master' into query_profiler
Conflicts:
core/src/main/java/org/elasticsearch/search/SearchService.java
commit 9b29d6823a71140ecd872df25ff9f7478e7fe766
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Dec 14 16:13:23 2015 -0500
[TEST] Profile flag needs to be set, ensure searches go against primary only for consistency
commit 4d602d8ad1f8cbc7b475450921fa3bc7d395b34f
Merge: 8b48e87 7742c1e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Dec 14 10:56:25 2015 -0500
Merge remote-tracking branch 'upstream/master' into query_profiler
commit 8b48e876348b163ab730eeca7fa35142165b05f9
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Dec 14 10:56:01 2015 -0500
Delegate straight to in.matchCost, no need for profiling
commit fde3b0587911f0b5f15e779c671d0510cbd568a9
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Dec 14 10:28:23 2015 -0500
Documentation tweaks, renaming build_weight -> create_weight
commit 46f5e011ee23fe9bb8a1f11ceb4fa9d19fe48e2e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Dec 14 10:27:52 2015 -0500
Profile TwoPhaseIterator should override matchCost()
commit b59f894ddb11b2a7beebba06c4ec5583ff91a7b2
Merge: 9aa1a3a b4e0c87
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 9 14:23:26 2015 -0500
Merge remote-tracking branch 'upstream/master' into query_profiler
commit 9aa1a3a25c34c9cd9fffaa6114c25a0ec791307d
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 9 13:41:48 2015 -0500
Revert "Move some of the collector wrapping logic into ProfileCollectorBuilder"
This reverts commit 02cc31767fb76a7ecd44a302435e93a05fb4220e.
commit 57f7c04cea66b3f98ba2bec4879b98e4fba0b3c0
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 9 13:41:31 2015 -0500
Revert "Rearrange if/else to make intent clearer"
This reverts commit 59b63c533fcaddcdfe4656e86a6f6c4cb1bc4a00.
commit 2874791b9c9cd807113e75e38be465f3785c154e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 9 13:38:13 2015 -0500
Revert "Move state into ProfileCollectorBuilder"
This reverts commit 0bb3ee0dd96170b06f07ec9e2435423d686a5ae6.
commit 0bb3ee0dd96170b06f07ec9e2435423d686a5ae6
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Dec 3 11:21:55 2015 -0500
Move state into ProfileCollectorBuilder
commit 59b63c533fcaddcdfe4656e86a6f6c4cb1bc4a00
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 2 17:21:12 2015 -0500
Rearrange if/else to make intent clearer
commit 511db0af2f3a86328028b88a6b25fa3dfbab963b
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 2 17:12:06 2015 -0500
Rename WEIGHT -> BUILD_WEIGHT
commit 02cc31767fb76a7ecd44a302435e93a05fb4220e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Dec 2 17:11:22 2015 -0500
Move some of the collector wrapping logic into ProfileCollectorBuilder
commit e69356d3cb4c60fa281dad36d84faa64f5c32bc4
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 15:12:35 2015 -0500
Cleanup imports
commit c1b4f284f16712be60cd881f7e4a3e8175667d62
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 15:11:25 2015 -0500
Review cleanup: Make InternalProfileShardResults writeable
commit 9e61c72f7e1787540f511777050a572b7d297636
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 15:01:22 2015 -0500
Review cleanup: Merge ProfileShardResult, InternalProfileShardResult. Convert to writeable
commit 709184e1554f567c645690250131afe8568a5799
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 14:38:08 2015 -0500
Review cleanup: Merge ProfileResult, InternalProfileResult. Convert to writeable
commit 7d72690c44f626c34e9c608754bc7843dd7fd8fe
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 14:01:34 2015 -0500
Review cleanup: use primitive (and default) for profile flag
commit 97d557388541bbd3388cdcce7d9718914d88de6d
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 13:09:12 2015 -0500
Review cleanup: Use Collections.emptyMap() instead of building an empty one explicitly
commit 219585b8729a8b0982e653d99eb959efd0bef84e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 13:08:12 2015 -0500
Add todo to revisit profiler architecture in the future
commit b712edb2160e032ee4b2f2630fadf131a0936886
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 30 13:05:32 2015 -0500
Split collector serialization from timing, use writeable instead of streamable
Previously, the collector timing was done in the same class that was serialized, which required
leaving the collector null when serializing. Besides being a bit gross, this made it difficult to
change the class to Writeable.
This splits up the timing (InternalProfileCollector + ProfileCollector) and the serialization of
the times (CollectorResult). CollectorResult is writeable, and also acts as the public interface.
commit 6ddd77d066262d4400e3d338b11cebe7dd27ca78
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Nov 25 13:15:12 2015 -0500
Remove dead code
commit 06033f8a056e2121d157654a65895c82bbe93a51
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Nov 25 12:49:51 2015 -0500
Review cleanup: Delegate to in.getProfilers()
Note: Need to investigate how this is used exactly so we can add a test, it isn't touched by a
normal inner_hits query...
commit a77e13da21b4bad1176ca2b5d5b76034fb12802f
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Nov 25 11:59:58 2015 -0500
Review cleanup: collapse to single `if` statement
commit e97bb6215a5ebb508b0293ac3acd60d5ae479be1
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Nov 25 11:39:43 2015 -0500
Review cleanup: Return empty map instead of null for profile results
Note: we still need to check for nullness in SearchPhaseController, since an empty/no-hits result
won't have profiling instantiated (or any other component like aggs or suggest). Therefore
QuerySearchResult.profileResults() is still @Nullable
commit db8e691de2a727389378b459fa76c942572e6015
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Nov 25 10:14:47 2015 -0500
Review cleanup: renaming, formatting fixes, assertions
commit 9011775fe80ba22c2fd948ca64df634b4e32772d
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Nov 19 20:09:52 2015 -0500
[DOCS] Add missing annotation
commit 4b58560b06f08d4b99b149af20916ee839baabd7
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Nov 19 20:07:17 2015 -0500
[DOCS] Update documentation for new format
commit f0458c58e5538ed8ec94849d4baf3250aa9ec841
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Nov 17 10:14:09 2015 +0100
Reduce visibility of internal classes.
commit d0a7d319098e60b028fa772bf8a99b2df9cf6146
Merge: e158070 1bdf29e
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Nov 17 10:09:18 2015 +0100
Merge branch 'master' into query_profiler
commit e158070a48cb096551f3bb3ecdcf2b53bbc5e3c5
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Nov 17 10:08:48 2015 +0100
Fix compile error due to bad html in javadocs.
commit a566b5d08d659daccb087a9afbe908ec3d96cd6e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 16 17:48:37 2015 -0500
Remove unused collector
commit 4060cd72d150cc68573dbde62ca7321c47f75703
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 16 17:48:10 2015 -0500
Comment cleanup
commit 43137952bf74728f5f5d5a8d1bfc073e0f9fe4f9
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Nov 16 17:32:06 2015 -0500
Fix negative formatted time
commit 5ef3a980266326aff12d4fe380f73455ff28209f
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 17:10:17 2015 +0100
Fix javadocs.
commit 276114d29e4b17a0cc0982cfff51434f712dc59e
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 16:25:23 2015 +0100
Fix: include rewrite time as well...
commit 21d9e17d05487bf4903ae3d2ab6f429bece2ffef
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 15:10:15 2015 +0100
Remove TODO about profiling explain.
commit 105a31e8e570efb879447159c3852871f5cf7db4
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 14:59:30 2015 +0100
Fix nocommit now that the global collector is a root collector.
commit 2e8fc5cf84adb1bfaba296808c329e5f982c9635
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 14:53:38 2015 +0100
Make collector wrapping more explicit/robust (and a bit less magical).
commit 5e30b570b0835e1ce79a57933a31b6a2d0d58e2d
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 12:44:03 2015 +0100
Simplify recording API a bit.
commit 9b453afced6adc0a59ca1d67d90c28796b105185
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 10:54:25 2015 +0100
Fix serialization-related nocommits.
commit ad97b200bb123d4e9255e7c8e02f7e43804057a5
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Nov 13 10:46:30 2015 +0100
Fix DFS.
commit a6de06986cd348a831bd45e4f524d2e14d9e03c3
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Nov 12 19:29:16 2015 +0100
Remove forbidden @Test annotation.
commit 4991a28e19501109af98026e14756cb25a56f4f4
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Nov 12 19:25:59 2015 +0100
Limit the impact of query profiling on the SearchContext API.
Rule is: we can put as much as we want in the search.profile package but should
aim at touching as little as possible other areas of the code base.
commit 353d8d75a5ce04d9c3908a0a63d4ca6e884c519a
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Nov 12 18:05:09 2015 +0100
Remove dead code.
commit a3ffafb5ddbb5a2acf43403c946e5ed128f47528
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Nov 12 15:30:35 2015 +0100
Remove call to forbidden String.toLowerCase() API.
commit 1fa8c7a00324fa4e32bd24135ebba5ecf07606f1
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Nov 12 15:30:27 2015 +0100
Fix compilation.
commit 2067f1797e53bef0e1a8c9268956bc5fb8f8ad97
Merge: 22e631f fac472f
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Nov 12 15:21:12 2015 +0100
Merge branch 'master' into query_profiler
commit 22e631fe6471fed19236578e97c628d5cda401a9
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Nov 3 18:52:05 2015 -0500
Fix and simplify serialization of shard profile results
commit 461da250809451cd2b47daf647343afbb4b327f2
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Nov 3 18:32:22 2015 -0500
Remove helper methods, simpler without them
commit 5687aa1c93d45416d895c2eecc0e6a6b302139f2
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Nov 3 18:29:32 2015 -0500
[TESTS] Fix tests for new rewrite format
commit ba9e82857fc6d4c7b72ef4d962d2102459365299
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 30 15:28:14 2015 -0400
Rewrites begone! Record all rewrites as a single time metric
commit 5f28d7cdff9ee736651d564f71f713bf45fb1d91
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 29 15:36:06 2015 -0400
Merge duplicate rewrites into one entry
By using the Query as the key in a map, we can easily merge rewrites together. This means
the preProcess(), assertion and main query rewrites all get merged together. Downside is that
rewrites of the same Query (hashcode) but in different places get lumped together. I think the
simplicity of the solution is better than the slight loss in output fidelity.
commit 9a601ea46bb21052746157a45dcc6de6bc350e9c
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 29 15:28:27 2015 -0400
Allow multiple "searches" per profile (e.g. query + global agg)
commit ee30217328381cd83f9e653d3a4d870c1d2bdfce
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 29 11:29:18 2015 -0400
Update comment, add nullable annotation
commit 405c6463a64e118f170959827931e8c6a1661f13
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 29 11:04:30 2015 -0400
remove out-dated comment
commit 2819ae8f4cf1bfd5670dbd1c0e06195ae457b58f
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Oct 27 19:50:47 2015 +0100
Don't render children in the profiles when there are no children.
commit 7677c2ddefef321bbe74660471603d202a4ab66f
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Oct 27 19:50:35 2015 +0100
Set the profiler on the ContextIndexSearcher.
commit 74a4338c35dfed779adc025ec17cfd4d1c9f66f5
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Oct 27 19:50:01 2015 +0100
Fix json rendering.
commit 6674d5bebe187b0b0d8b424797606fdf2617dd27
Author: Adrien Grand <jpountz@gmail.com>
Date: Tue Oct 27 19:20:19 2015 +0100
Revert "nocommit - profiling never enabled because setProfile() on ContextIndexSearcher never called"
This reverts commit d3dc10949024342055f0d4fb7e16c7a43423bfab.
commit d3dc10949024342055f0d4fb7e16c7a43423bfab
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 17:20:57 2015 -0400
nocommit - profiling never enabled because setProfile() on ContextIndexSearcher never called
Previously, it was enabled by using DefaultSearchContext as a third-party "proxy", but since
the refactor to make it unit testable, setProfile() needs to be called explicitly. Unfortunately,
this is not possible because SearchService only has access to an IndexSearcher. And it is not
cast'able to a DefaultIndexSearcher.
commit b9ba9c5d1f93b9c45e97b0a4e35da6f472c9ea53
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 16:27:00 2015 -0400
[TESTS] Fix unit tests
commit cf5d1e016b2b4a583175e07c16c7152f167695ce
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 16:22:34 2015 -0400
Increment token after recording a rewrite
commit b7d08f64034e498533c4a81bff8727dd8ac2843e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 16:14:09 2015 -0400
Fix NPE if a top-level root doesn't have children
commit e4d3b514bafe2a3a9db08438c89f0ed68628f2d6
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 16:05:47 2015 -0400
Fix NPE when profiling is disabled
commit 445384fe48ed62fdd01f7fc9bf3e8361796d9593
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 16:05:37 2015 -0400
[TESTS] Fix integration tests
commit b478296bb04fece827a169e7522df0a5ea7840a3
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 15:43:24 2015 -0400
Move rewrites to their own section, remove reconciliation
Big commit because the structural change affected a lot of the wrapping code. Major changes:
- Rewrites are now in their own section in the response
- Reconciliation is gone...we don't attempt to roll the rewrites into the query tree structure
- InternalProfileShardResults (plural) simply holds a Map<String, InternalProfileShardResult> and
helps to serialize / ToXContent
- InternalProfileShardResult (singular) is now the holder for all shard-level profiling details. Currently
this includes query, collectors and rewrite. In the future it would hold suggest, aggs, etc
- When the user requests the profiled results, they get back a Map<String, ProfileShardResult>
instead of doing silly helper methods to convert to maps, etc
- Shard details are baked into a string instead of serializing the object
commit 24819ad094b208d0e94f17ce9c3f7c92f7414124
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Oct 23 10:25:38 2015 -0400
Make Profile results immutable by removing relative_time
commit bfaf095f45fed74194ef78160a8e5dcae1850f9e
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 23 10:54:59 2015 +0200
Add nocommits.
commit e9a128d0d26d5b383b52135ca886f2c987850179
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 23 10:39:37 2015 +0200
Move all profile-related classes to the same package.
commit f20b7c7fdf85384ecc37701bb65310fb8c20844f
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Oct 23 10:33:14 2015 +0200
Reorganize code a bit to ease unit testing of ProfileCollector.
commit 3261306edad6a0c70f59eaee8fe58560f61a75fd
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 18:07:28 2015 +0200
Remove irrelevant nocommit.
commit a6ac868dad12a2e17929878681f66dbd0948d322
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 18:06:45 2015 +0200
Make CollectorResult's reason a free-text field to ease bw compat.
commit 5d0bf170781a950d08b81871cd1e403e49f3cc12
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 16:50:52 2015 +0200
Add unit tests for ProfileWeight/ProfileScorer.
commit 2cd88c412c6e62252504ef69a59216adbb574ce4
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 15:55:17 2015 +0200
Rename InternalQueryProfiler to Profiler.
commit 84f5718fa6779f710da129d9e0e6ff914fd85e36
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 15:53:58 2015 +0200
Merge InternalProfileBreakdown into ProfileBreakdown.
commit 135168eaeb8999c8117ea25288104b0961ce9b35
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 13:56:57 2015 +0200
Make it possible to instantiate a ContextIndexSearcher without SearchContext.
commit 5493fb52376b48460c4ce2dedbe00cc5f6620499
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 11:53:29 2015 +0200
Move ProfileWeight/Scorer to their own files.
commit bf2d917b9dae3b32dfc29c35a7cac4ccb7556cce
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 11:38:24 2015 +0200
Fix bug that caused phrase queries to fail.
commit b2bb0c92c343334ec1703a221af24a1b55e36d53
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 11:36:17 2015 +0200
Parsing happens on the coordinating node now.
commit 416cabb8621acb5cd8dfa77374fd23e428f52fe9
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 11:22:17 2015 +0200
Fix compilation (in particular remove guava deps).
commit f996508645f842629d403fc2e71c1890c0e2cac9
Merge: 4616a25 bc3b91e
Author: Adrien Grand <jpountz@gmail.com>
Date: Thu Oct 22 10:44:38 2015 +0200
Merge branch 'master' into query_profiler
commit 4616a25afffe9c24c6531028f7fccca4303d2893
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Oct 20 12:11:32 2015 -0400
Make Java Count API compatible with profiling
commit cbfba74e16083d719722500ac226efdb5cb2ff55
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Oct 20 12:11:19 2015 -0400
Fix serialization of profile query param, NPE
commit e33ffac383b03247046913da78c8a27e457fae78
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Oct 20 11:17:48 2015 -0400
TestSearchContext should return null Profiler instead of exception
commit 73a02d69b466dc1a5b8a5f022464d6c99e6c2ac3
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Oct 19 12:07:29 2015 -0400
[DOCS] Update docs to reflect new ID format
commit 36248e388c354f954349ecd498db7b66f84ce813
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Oct 19 12:03:03 2015 -0400
Use the full [node][index][shard] string as profile result ID
commit 5cfcc4a6a6b0bcd6ebaa7c8a2d0acc32529a80e1
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 15 17:51:40 2015 -0400
Add failing test for phrase matching
Stack trace generated:
[2015-10-15 17:50:54,438][ERROR][org.elasticsearch.search.profile] shard [[JNj7RX_oSJikcnX72aGBoA][test][2]], reason [RemoteTransportException[[node_s0][local[1]][indices:data/read/search[phase/query]]]; nested: QueryPhaseExecutionException[Query Failed [Failed to execute main query]]; nested: AssertionError[nextPosition() called more than freq() times!]; ], cause [java.lang.AssertionError: nextPosition() called more than freq() times!
at org.apache.lucene.index.AssertingLeafReader$AssertingPostingsEnum.nextPosition(AssertingLeafReader.java:353)
at org.apache.lucene.search.ExactPhraseScorer.phraseFreq(ExactPhraseScorer.java:132)
at org.apache.lucene.search.ExactPhraseScorer.access$000(ExactPhraseScorer.java:27)
at org.apache.lucene.search.ExactPhraseScorer$1.matches(ExactPhraseScorer.java:69)
at org.elasticsearch.common.lucene.search.ProfileQuery$ProfileScorer$2.matches(ProfileQuery.java:226)
at org.apache.lucene.search.ConjunctionDISI$TwoPhaseConjunctionDISI.matches(ConjunctionDISI.java:175)
at org.apache.lucene.search.ConjunctionDISI$TwoPhase.matches(ConjunctionDISI.java:213)
at org.apache.lucene.search.ConjunctionDISI.doNext(ConjunctionDISI.java:128)
at org.apache.lucene.search.ConjunctionDISI.nextDoc(ConjunctionDISI.java:151)
at org.apache.lucene.search.ConjunctionScorer.nextDoc(ConjunctionScorer.java:62)
at org.elasticsearch.common.lucene.search.ProfileQuery$ProfileScorer$1.nextDoc(ProfileQuery.java:205)
at org.apache.lucene.search.Weight$DefaultBulkScorer.scoreAll(Weight.java:224)
at org.apache.lucene.search.Weight$DefaultBulkScorer.score(Weight.java:169)
at org.apache.lucene.search.BulkScorer.score(BulkScorer.java:39)
at org.apache.lucene.search.IndexSearcher.search(IndexSearcher.java:795)
at org.apache.lucene.search.IndexSearcher.search(IndexSearcher.java:509)
at org.elasticsearch.search.query.QueryPhase.execute(QueryPhase.java:347)
at org.elasticsearch.search.query.QueryPhase.execute(QueryPhase.java:111)
at org.elasticsearch.search.SearchService.loadOrExecuteQueryPhase(SearchService.java:366)
at org.elasticsearch.search.SearchService.executeQueryPhase(SearchService.java:378)
at org.elasticsearch.search.action.SearchServiceTransportAction$SearchQueryTransportHandler.messageReceived(SearchServiceTransportAction.java:368)
at org.elasticsearch.search.action.SearchServiceTransportAction$SearchQueryTransportHandler.messageReceived(SearchServiceTransportAction.java:365)
at org.elasticsearch.transport.local.LocalTransport$2.doRun(LocalTransport.java:280)
at org.elasticsearch.common.util.concurrent.AbstractRunnable.run(AbstractRunnable.java:37)
at java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
at java.lang.Thread.run(Thread.java:745)
commit 889fe6383370fe919aaa9f0af398e3040209e40b
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 15 17:30:38 2015 -0400
[DOCS] More docs
commit 89177965d031d84937753538b88ea5ebae2956b0
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Oct 15 09:59:09 2015 -0400
Fix multi-stage rewrites to recursively find most appropriate descendant rewrite
Previously, we chose the first rewrite that matched. But in situations where a query may
rewrite several times, this won't build the tree correctly. Instead we need to recurse
down all the rewrites until we find the most appropriate "leaf" rewrite
The implementation of this is kinda gross: we recursively call getRewrittenParentToken(),
which does a linear scan over the rewriteMap and tries to find rewrites with a larger token
value (since we know child tokens are always larger). Can almost certainly find a better
way to do this...
commit 0b4d782b5348e5d03fd26f7d91bc4a3fbcb7f6a5
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Oct 14 19:30:06 2015 -0400
[Docs] Documentation checkpoint
commit 383636453f6610fcfef9070c21ae7ca11346793e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Sep 16 16:02:22 2015 -0400
Comments
commit a81e8f31e681be16e89ceab9ba3c3e0a018f18ef
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Sep 16 15:48:49 2015 -0400
[TESTS] Ensure all tests use QUERY_THEN_FETCH, DFS does not profile
commit 1255c2d790d85fcb9cbb78bf2a53195138c6bc24
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Sep 15 16:43:46 2015 -0400
Refactor rewrite handling to handle identical rewrites
commit 85b7ec82eb0b26a6fe87266b38f5f86f9ac0c44f
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Sep 15 08:51:14 2015 -0400
Don't update parent when a token is added as root -- Fixes NPE
commit 109d02bdbc49741a3b61e8624521669b0968b839
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Sep 15 08:50:40 2015 -0400
Don't set the rewritten query if not profiling -- Fixes NPE
commit 233cf5e85f6f2c39ed0a2a33d7edd3bbd40856e8
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Sep 14 18:04:51 2015 -0400
Update tests to new response format
commit a930b1fc19de3a329abc8ffddc6711c1246a4b15
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Sep 14 18:03:58 2015 -0400
Fix serialization
commit 69afdd303660510c597df9bada5531b19d134f3d
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Sep 14 15:11:31 2015 -0400
Comments and cleanup
commit 64e7ca7f78187875378382ec5d5aa2462ff71df5
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Sep 14 14:40:21 2015 -0400
Move timing into dedicated class, add proper rewrite integration
commit b44ff85ddbba0a080e65f2e7cc8c50d30e95df8e
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Sep 14 12:00:38 2015 -0400
Checkpoint - Refactoring to use a token-based dependency tree
commit 52cedd5266d6a87445c6a4cff3be8ff2087cd1b7
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Fri Sep 4 19:18:19 2015 -0400
Need to set context profiling flag before calling queryPhase.preProcess
commit c524670cb1ce29b4b3a531fa2bff0c403b756f46
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Sep 4 18:00:37 2015 +0200
Reduce profiling overhead a bit.
This removes hash-table lookups everytime we start/stop a profiling clock.
commit 111444ff8418737082236492b37321fc96041e09
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Sep 4 16:18:59 2015 +0200
Add profiling of two-phase iterators.
This is useful for eg. phrase queries or script filters, since they are
typically consumed through their two-phase iterator instead of the scorer.
commit f275e690459e73211bc8494c6de595c0320f4c0b
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Sep 4 16:03:21 2015 +0200
Some more improvements.
I changed profiling to disable bulk scoring, since it makes it impossible to
know where time is spent. Also I removed profiling of operations that are
always fast (eg. normalization) and added nextDoc/advance.
commit 3c8dcd872744de8fd76ce13b6f18f36f8de44068
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Sep 4 14:39:50 2015 +0200
Remove println.
commit d68304862fb38a3823aebed35a263bd9e2176c2f
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Sep 4 14:36:03 2015 +0200
Fix some test failures introduced by the rebase...
commit 04d53ca89fb34b7a21515d770c32aaffcc513b90
Author: Adrien Grand <jpountz@gmail.com>
Date: Fri Sep 4 13:57:35 2015 +0200
Reconcile conflicting changes after rebase
commit fed03ec8e2989a0678685cd6c50a566cec42ea4f
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Thu Aug 20 22:40:39 2015 -0400
Add Collectors to profile results
Profile response element has now been re-arranged so that everything is listed per-shard to
facilitate grouping elements together. The new `collector` element looks like this:
```
"profile": {
"shards": [
{
"shard_id": "keP4YFywSXWALCl4m4k24Q",
"query": [...],
"collector": [
{
"name": "MultiCollector",
"purpose": "search_multi",
"time": "16.44504400ms",
"relative_time": "100.0000000%",
"children": [
{
"name": "FilteredCollector",
"purpose": "search_post_filter",
"time": "4.556013000ms",
"relative_time": "27.70447437%",
"children": [
{
"name": "SimpleTopScoreDocCollector",
"purpose": "search_sorted",
"time": "1.352166000ms",
"relative_time": "8.222331299%",
"children": []
}
]
},
{
"name": "BucketCollector: [[non_global_term, another_agg]]",
"purpose": "aggregation",
"time": "10.40379400ms",
"relative_time": "63.26400829%",
"children": []
},
...
```
commit 1368b495c934be642c00f6cbf9fc875d7e6c07ff
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Aug 19 12:43:03 2015 -0400
Move InternalProfiler to profile package
commit 53584de910db6d4a6bb374c9ebb954f204882996
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 18:34:58 2015 -0400
Only reconcile rewrite timing when rewritten query is different from original
commit 9804c3b29d2107cd97f1c7e34d77171b62cb33d0
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 16:40:15 2015 -0400
Comments and cleanup
commit 8e898cc7c59c0c1cc5ed576dfed8e3034ca0967f
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 14:19:07 2015 -0400
[TESTS] Fix comparison test to ensure results sort identically
commit f402a29001933eef29d5a62e81c8563f1c8d0969
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 14:17:59 2015 -0400
Add note about DFS being unable to profile
commit d446e08d3bc91cd85b24fc908e2d82fc5739d598
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 14:17:23 2015 -0400
Implement some missing methods
commit 13ca94fb86fb037a30d181b73d9296153a63d6e4
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 13:10:54 2015 -0400
[TESTS] Comments & cleanup
commit c76c8c771fdeee807761c25938a642612a6ed8e7
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 13:06:08 2015 -0400
[TESTS] Fix profileMatchesRegular to handle NaN scores and nearlyEqual floats
commit 7e7a10ecd26677b2239149468e24938ce5cc18e1
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 12:22:16 2015 -0400
Move nearlyEquals() utility function to shared location
commit 842222900095df4b27ff3593dbb55a42549f2697
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 18 12:04:35 2015 -0400
Fixup rebase conflicts
commit 674f162d7704dd2034b8361358decdefce1f76ce
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Aug 17 15:29:35 2015 -0400
[TESTS] Update match and bool tests
commit 520380a85456d7137734aed0b06a740e18c9cdec
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Aug 17 15:28:09 2015 -0400
Make naming consistent re: plural
commit b9221501d839bb24d6db575d08e9bee34043fc65
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Aug 17 15:27:39 2015 -0400
Children need to be added to list after serialization
commit 05fa51df940c332fbc140517ee56e849f2d40a72
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Aug 17 15:22:41 2015 -0400
Re-enable bypass for non-profiled queries
commit f132204d264af77a75bd26a02d4e251a19eb411d
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Aug 17 15:21:14 2015 -0400
Fix serialization of QuerySearchResult, InternalProfileResult
commit 27b98fd475fc2e9508c91436ef30624bdbee54ba
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Aug 10 17:39:17 2015 -0400
Start to add back tests, refactor Java api
commit bcfc9fefd49307045108408dc160774666510e85
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Aug 4 17:08:10 2015 -0400
Checkpoint
commit 26a530e0101ce252450eb23e746e48c2fd1bfcae
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Tue Jul 14 13:30:32 2015 -0400
Add createWeight() checkpoint
commit f0dd61de809c5c13682aa213c0be65972537a0df
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Mon Jul 13 12:36:27 2015 -0400
checkpoint
commit 377ee8ce5729b8d388c4719913b48fae77a16686
Author: Zachary Tong <zacharyjtong@gmail.com>
Date: Wed Mar 18 10:45:01 2015 -0400
checkpoint
2015-12-17 12:05:43 -05:00
|
|
|
public Profilers getProfilers() {
|
|
|
|
return profilers;
|
|
|
|
}
|
|
|
|
|
|
|
|
public void setProfilers(Profilers profilers) {
|
|
|
|
this.profilers = profilers;
|
|
|
|
}
|
2016-10-25 12:27:33 -10:00
|
|
|
|
|
|
|
@Override
|
2019-11-28 18:23:09 +01:00
|
|
|
public void setTask(SearchShardTask task) {
|
2016-10-25 12:27:33 -10:00
|
|
|
this.task = task;
|
|
|
|
}
|
|
|
|
|
|
|
|
@Override
|
2019-11-28 18:23:09 +01:00
|
|
|
public SearchShardTask getTask() {
|
2016-10-25 12:27:33 -10:00
|
|
|
return task;
|
|
|
|
}
|
|
|
|
|
|
|
|
@Override
|
|
|
|
public boolean isCancelled() {
|
|
|
|
return task.isCancelled();
|
|
|
|
}
|
2013-08-26 16:28:27 +02:00
|
|
|
}
|