Commit Graph

23688 Commits

Author SHA1 Message Date
Alexander Lin 119026b4fb Remove isCreated and isFound from the Java API
This is cleanup work from #19566, where @nik9000 suggested trying to nuke the isCreated and isFound methods. I've combined nuking the two methods with removing UpdateHelper.Operation in favor of DocWriteResponse.Operation here.

Closes #19631.
2016-07-29 14:21:43 -04:00
Nik Everett c9790a1257 Use fewer threads when reindexing-from-remote
Reindex from remote uses the Elasticsearch client which uses apache
httpasyncclient which spins up 5 thread by default, 1 as a dispatcher
and 4 more to handle IO. This changes Reindex's usage so it only spins
up two thread - 1 dispatcher and one to handle io. It also renames the
threads to "es-client-$taskid-$thread_number". That way if we see any
thread sticking around we can trace it back to the task.
2016-07-29 14:13:10 -04:00
Christoph Büscher 4450039cf6 Try catching potential null query results and convert to MatchNoDocsQuery 2016-07-29 18:29:48 +02:00
Martijn van Groningen a91bb29585 ingest: Made the response format of the get pipeline api match with the response format of the index template api
Closes #19585
2016-07-29 17:58:30 +02:00
Alexander Reelsen 30b497622b Dependencies: Upgrade to netty 4.1.4 (#19689)
Removes our internal snapshot repo we used until 4.1.4 was released.
2016-07-29 17:53:12 +02:00
Mike McCandless 37e0e63a65 add defense to selectNewPathForShard 2016-07-29 11:51:33 -04:00
Ryan Ernst 05edb0367d Add client jars pom generation, and make transport client depend on
transport client plugin jars
2016-07-29 07:41:39 -07:00
Nik Everett ad028f3f9c Squash o.e.rest.action.admin.indices
In an effort to reduce the number of tiny packages we have in the
code base this moves all the files that were in subdirectories of
`org.elasticsearch.rest.action.admin.indices` into
`org.elasticsearch.rest.action.admin.indices`.

It also adds a `package-info.java` file explaining what the files in
the package *do*.

Also fixes line length in these packages. It makes a single non-checkstyle
change: implementing `ToXContent` on `GetIndexTemplatesResponse`. I did
this because it was the right thing to do and it fixed a line length
violation.
2016-07-29 10:08:03 -04:00
Colin Goodheart-Smithe 3f13f02575 [DOCS] updated documentation for transport client changes
Updated dependency in Java API docs and added section in breaking
changes
2016-07-29 14:25:12 +01:00
Martijn van Groningen 7b36a72ccb fixed compiler warnining and removed unused imports 2016-07-29 15:06:49 +02:00
Martijn van Groningen 81112508ea test: fix type in test name 2016-07-29 14:52:24 +02:00
Clinton Gormley f92b3c965e Rename "breaking java" tag to "breaking-java" 2016-07-29 14:43:31 +02:00
Clinton Gormley 1cbe3bf2ad Updated release notes to use breaking-java tag 2016-07-29 14:43:31 +02:00
Clinton Gormley 3922392218 Added release notes for 5.0.0-alpha5 2016-07-29 14:43:31 +02:00
Clinton Gormley e1a347bed5 Update release notes script to add the "breaking java" label 2016-07-29 14:43:31 +02:00
David Pilato 6b68d1e09b Fix typo in comment 2016-07-29 13:49:11 +02:00
Martijn van Groningen 72e0d422e9 Plain highlighter should ignore parent/child queries.
The plain highligher fails when it tries to select the fragments based on a query containing either a `has_child` or `has_parent` query.

The plain highligher should just ignore parent/child queries as it makes no sense to highligh a parent match with a has_child as the child documents are not available at highlight time. Instead if child document should be highlighed inner hits should be used.

Parent/child queries already have no effect when the `fvh` or `postings` highligher is used. The test added in this commit verifies that.

Closes #14999
2016-07-29 12:41:11 +02:00
Luca Cavanna 502217f035 [DOCS] add java REST client docs (#19618)
[DOCS] add java REST client docs

Add some docs on how to get started with the Java REST client, some common configuration that may be needed and the sniffer component.
2016-07-29 11:22:47 +02:00
Christoph Büscher 757de805d3 `multi_match` query should produce MatchNoDocs query on unknown fieldname
Currently when the `fields` parameter used in a `multi_match` query contains a
wildcard expression that doesn't resolve to any field name in the target index,
MultiMatchQueryBuilder produces a `null` query. This change changes it to be a
MatchNoDocs query, since returning no documents for this case is already the
current behaviour. Also adding missing field names (with and without wildcards)
to the unit and integration test.
2016-07-29 10:56:37 +02:00
Colin Goodheart-Smithe 5a99ce5b91 #19676 Added JavaDocs and comments to ParseField
Added JavaDocs and comments to ParseField
2016-07-29 09:49:40 +01:00
Colin Goodheart-Smithe f1257bfb86 Added JavaDocs and comments to ParseField 2016-07-29 09:39:38 +01:00
Colin Goodheart-Smithe cd88b7724e Undeprecates `aggs` in the search request
This change adds a second ParseField for the `aggs` field in the search
request so both `aggregations` and `aggs` are undeprecated allowed
fields in the search request

Closes #19504
2016-07-29 09:14:32 +01:00
Adrien Grand dcc598c414 Make the heuristic to compute the default shard size less aggressive.
The current heuristic to compute a default shard size is pretty aggressive,
it returns `max(10, number_of_shards * size)` as a value for the shard size.
I think making it less aggressive has the benefit that it would reduce the
likelyness of running into OOME when there are many shards (yearly
aggregations with time-based indices can make numbers of shards in the
thousands) and make the use of breadth-first more likely/efficient.

This commit replaces the heuristic with `size * 1.5 + 10`, which is enough
to have good accuracy on zipfian distributions.
2016-07-29 09:59:29 +02:00
Ali Beyad 58d6b9dcd1 This commit first reads the repository data and only
upgrades if it determines the read data is in the legacy
format. It writes the upgraded version if it is not a
read-only repository and caches the repository data if
it is a read-only repository.
2016-07-28 22:09:01 -04:00
Brandon Wulf 6b7d40929c Switch example from inclusion to exclusion.
Page is explaining allocation exclusion- example should be about exclusion as well.
2016-07-28 21:54:22 -04:00
Nik Everett e04f06258f Assert we return Location header with 201 CREATED
Add an assertion to the most popular way of turning the response object
into the actual http response. As it stands all places we return
`201 CREATED` we return the `Location` header. This will help to keep it
that way, though it won't catch all uses.

Followup to #19509
2016-07-28 16:13:58 -04:00
Nik Everett 2e7336dc10 Add package-info to o.e.test.rest
This removes two packages, consolidating them into their parent package
and adds `package-info.java` files to describe all of the packages under
`org.elasticsearch.test.rest`.
2016-07-28 16:07:44 -04:00
Ryan Ernst c414a6cedd Build: Fix client jars for plugins/modules to have the correct artifactId
For transport client plugins, the jars and poms are renamed with the
-client suffix. But we need the artifactId to match the id in the
filename.
2016-07-28 12:19:10 -07:00
David Pilato f8e0557be5 Extract AWS Key from KeyChain instead of using potential null value
While I was working on #18703, I discovered a bad behavior when people don't provide AWS key/secret as part as their `elasticsearch.yml` but rely on SysProps or env. variables...

In [`InternalAwsS3Service#getClient(...)`](d4366f8493/plugins/repository-s3/src/main/java/org/elasticsearch/cloud/aws/InternalAwsS3Service.java (L76-L141)), we have:

```java
        Tuple<String, String> clientDescriptor = new Tuple<>(endpoint, account);
        AmazonS3Client client = clients.get(clientDescriptor);
```

But if people don't provide credentials, `account` is `null`.

Even if it actually could work, I think that we should use the `AWSCredentialsProvider` we create later on and extract from it the `account` (AWS KEY actually) and then use it as the second value of the tuple.

Closes #19557.
2016-07-28 18:05:51 +02:00
Mike McCandless ed5e5db188 merge master 2016-07-28 11:55:16 -04:00
David Pilato fef0ffc49e Merge remote-tracking branch 'dadoonet/pr/19556-use-DefaultAWSCredentialsProviderChain' 2016-07-28 17:51:38 +02:00
David Pilato e79492dc31 Merge remote-tracking branch 'origin/master' 2016-07-28 17:46:05 +02:00
Areek Zillur 69941931c7 Merge pull request #19610 from areek/enhancement/19484
Add zero-padding to auto-generated rollover index name increment
2016-07-28 11:44:50 -04:00
David Pilato 3adccd4560 Merge branch 'pr/19556-use-DefaultAWSCredentialsProviderChain' 2016-07-28 17:38:52 +02:00
David Pilato 72784f44f0 Remove non needed imports 2016-07-28 17:34:00 +02:00
Mike McCandless ef15e1b91f work around JDK bug: if FileStore.getXXXSpace APIs return negative value, change that to Long.MAX_VALUE instead 2016-07-28 11:31:16 -04:00
David Pilato 0d2ccf0989 Merge branch 'pr/15724-gce-network-host-master' 2016-07-28 16:59:18 +02:00
Nik Everett f159156931 [docs] Deprecate found and created (#19633)
These parts of delete and index response have been replaced with the
operation field.
2016-07-28 10:20:48 -04:00
Luca Cavanna 4d1cc5296c Merge pull request #19653 from javanna/enhancement/response_exception_body_message
Add response body to ResponseException error message
2016-07-28 15:53:14 +02:00
David Pilato 7b9ce1212f Merge branch 'fix/npe-simulate-pipeline-no-id' 2016-07-28 14:55:07 +02:00
javanna 1ea8f865d6 Add response body to ResponseException error message
Consuming the response body to make it part of the exception message means that it may not be readable anymore later, depending on whether the entity is repeatable or not. Turns out that the response body tells a lot about the error itself, and considering that we don't expect bodies to be incredibly big for errors, we can wrap the entity into a BufferedHttpEntity to make it repeatable.

Closes #19622
2016-07-28 14:44:20 +02:00
javanna c63c5dc6a9 another s/maven/gradle 2016-07-28 14:39:46 +02:00
adityasinghraghav 952bc928fc Update TESTING.asciidoc (#19652) 2016-07-28 14:38:13 +02:00
Colin Goodheart-Smithe bab3e766c7 #19649 Makes `m` case sensitive in TimeValue
Makes `m` case sensitive in TimeValue
2016-07-28 13:00:57 +01:00
Clinton Gormley 3c639f0673 Removed array-of-string example from search template
Relates to #19643
2016-07-28 13:49:53 +02:00
David Pilato d406b88857 Fix NPE when simulating a pipeline with no id
When you simulate a pipeline without specifying an id against a node where the request is redirected to a master node,
the request and the response is throwing a NPE:

```
java.lang.NullPointerException
    at __randomizedtesting.SeedInfo.seed([3B9536AC6AA23C06:DD62280CF765DA1F]:0)
    at org.elasticsearch.common.io.stream.StreamOutput.writeString(StreamOutput.java:300)
    at org.elasticsearch.action.ingest.SimulatePipelineRequest.writeTo(SimulatePipelineRequest.java:92)
    at org.elasticsearch.transport.local.LocalTransport.sendRequest(LocalTransport.java:222)
    at org.elasticsearch.test.transport.AssertingLocalTransport.sendRequest(AssertingLocalTransport.java:95)
    at org.elasticsearch.transport.TransportService.sendRequest(TransportService.java:470)
    at org.elasticsearch.action.TransportActionNodeProxy.execute(TransportActionNodeProxy.java:51)
    at org.elasticsearch.client.transport.support.TransportProxyClient.lambda$execute$441(TransportProxyClient.java:63)
    at org.elasticsearch.client.transport.TransportClientNodesService.execute(TransportClientNodesService.java:233)
    at org.elasticsearch.client.transport.support.TransportProxyClient.execute(TransportProxyClient.java:63)
    at org.elasticsearch.client.transport.TransportClient.doExecute(TransportClient.java:309)
    at org.elasticsearch.client.support.AbstractClient.execute(AbstractClient.java:403)
    at org.elasticsearch.client.FilterClient.doExecute(FilterClient.java:67)
    at org.elasticsearch.client.support.AbstractClient.execute(AbstractClient.java:403)
    at org.elasticsearch.client.support.AbstractClient$ClusterAdmin.execute(AbstractClient.java:710)
    at org.elasticsearch.action.ActionRequestBuilder.execute(ActionRequestBuilder.java:80)
    at org.elasticsearch.action.ActionRequestBuilder.execute(ActionRequestBuilder.java:54)
    at org.elasticsearch.action.ActionRequestBuilder.get(ActionRequestBuilder.java:62)
    at org.elasticsearch.ingest.bano.BanoProcessorIntegrationTest.testSimulateProcessorConfigTarget(BanoProcessorIntegrationTest.java:139)
```

This patch fixes this and adds some random tests.
2016-07-28 13:28:24 +02:00
Britta Weber 105dce0e07 fix explain in function_score if no function filter matches (#19185)
* fix explain in function_score if no function filter matches

When each function in function_score has a filter but none of them matches
we always assume 1 for the combined functions and then combine that with the
sub query score.
But the explanation did not reflect that because in case no function matched
we did not even use the actual score that was computed in the explanation.
2016-07-28 13:14:08 +02:00
markwalkom ebf96bbc35 Update gateway.asciidoc (#19572)
* Update gateway.asciidoc

Added a note to clarify that, in cases where nodes in a cluster have different setting, the node that is the elected master takes precedence over anything else.

* Update gateway.asciidoc

Updated as per @bleskes's comments
2016-07-28 13:09:05 +02:00
Colin Goodheart-Smithe eab5ceb9de Makes `m` case sensitive in TimeValue
The reason for this change is that currently if a user specifies e.g.`2M`
meaning 2 months as a time value instead of throwing an exception
explaining that time units in months are not supported (due to months
having variable time spans) we instead will parse this to 2 minutes.
This could be surprising to a user and could mean put a lot of load on
the cluster performing a task that was never intended and whose results
will be useless anyway.

It is generally accepted that `m` indicates minutes and `M` indicates
months with time values so this is consistent with the expectations a
user might have around specifying time units.

A concrete example of where this causes issues is in the decay score
function which uses TimeValue to parse the scale and offset parameters
of the decay into millisecond values to use in the calculation.

Relates to #19619
2016-07-28 11:27:24 +01:00
Lee Hinman 9fa33b6d07 [TEST] throw correct error within assertBusy in TruncateTranslogIT 2016-07-27 16:40:49 -06:00