Commit Graph

21373 Commits

Author SHA1 Message Date
Nik Everett a9e85182f1 Cut serial_diff and cumulative_sum to registerPipelineAggregation
and remove their PROTOTYPEs.

Relates to #17085
2016-04-18 08:44:49 -04:00
Nik Everett e9fa53b87f Cut global, missing, cardinality, and value_count to registerAggregation
and remove their PROTOTYPEs.

Related to #17085
2016-04-18 08:32:24 -04:00
Nik Everett 070b78517e Cut stats_bucket and extended_stats_bucket to registerPipelineAggregation
and remove their PROTOTYPE.

Relates to #17085
2016-04-18 08:19:08 -04:00
Jim Ferenczi 6e4273ae0b Merge pull request #17818 from jimferenczi/search_slow_log
Add missing index name to search slow log.
2016-04-18 12:40:53 +02:00
Danilo Vaz 2e2d8c1442 Updated copyright years to include 2016 (#17808) 2016-04-18 12:39:23 +02:00
Jim Ferenczi 3dc21e780b Add missing index name to search slow log.
This commits adds the index name as part of the logging message.
Closes #17025
2016-04-18 12:13:10 +02:00
Daniel Mitterdorfer b84a192884 Honor invariant Suggest.suggestions != null 2016-04-18 10:18:23 +02:00
Daniel Mitterdorfer b9faa13274 Balance request sizes in NettyHttpRequestSizeLimitIT
This test should demonstrate that a single (larger)
request is processed but on of multiple large concurrent requests
is rejected. This test broke too early under some circumstances in
network mode as the limit is quite low.

With this commit we reduce the size of the individual large
requests but issue more concurrent ones thus increasing stability
of this test.
2016-04-18 09:31:48 +02:00
Sergii Golubev 5ce3eb96b0 tophits-aggregation.asciidoc: fix a typo 2016-04-18 09:23:39 +02:00
Nik Everett d3b1306069 Reindex: never report negative throttled_until
Just clamp the value at 0. It isn't useful to tell the user "this
thread should have woken 5ms ago".

Closes #17783
2016-04-15 16:53:23 -04:00
Jim Ferenczi 4575d79522 Merge pull request #17776 from jimferenczi/simple_query_string_wildcard
Apply the default operator on analyzed wildcard in simple_query_string builder
2016-04-15 22:22:55 +02:00
Jim Ferenczi b4e6b63088 Apply the default operator on analyzed wildcard in simple_query_string builder:
* This is a followup from https://github.com/elastic/elasticsearch/pull/17711 where we now apply the default operator on analyzed wildcard query in query_string builder.
2016-04-15 21:56:50 +02:00
Nik Everett 65803f8abd Cut significant_terms to registerAggregation
and remove its PROTOTYPE.

Relates to #17085
2016-04-15 14:23:43 -04:00
Nik Everett d223090d7d Cut top_hits aggregation to registerAggregation
and remove its PROTOTYPE.

Relates to #17085
2016-04-15 14:15:12 -04:00
Nik Everett c94302d246 Cut geo aggregations to registerAggregation
and remove their prototypes.

Relates to #17085
2016-04-15 14:01:48 -04:00
Michael McCandless b55368b39d Merge pull request #17778 from mikemccand/imc_registered_settings
Switch to registered Settings for all IndexingMemoryController settings
2016-04-15 13:14:12 -04:00
Nik Everett c966d14f8b [build] Don't resolve the revision hash unless you need it
This fixes the build for folks that build without git installed locally
and should speed up the general case because we aren't trying to resolve
git information when it isn't really needed.
2016-04-15 11:35:05 -04:00
Alexander Reelsen 95579ca95a Build: Allow for file based deploy, sign packages
This allows for a local file based deploy without needed nexus
auth information.

Also signing of packages has been added, either via gradle.properties
or using system properties as a fallback.

The property build.repository allows to configure another endpoint if no
snapshot build is done.

Fix creation of .asc file for tar.gz distribution

Closes #17405
2016-04-15 17:14:05 +02:00
Nik Everett b5a58ece41 Cut nested and reverse_nested aggregations to registerAggregation
and remove their PROTOTYPES.

Relates to #17085
2016-04-15 10:19:03 -04:00
Adrien Grand eb4ba38032 Add points to SegmentStats. #17775
This way points memory and disk usage will be reported in the stats API.

Closes #16974
2016-04-15 15:58:26 +02:00
Nik Everett 7f6a765a1e Cut the sampler and diversified_sampler aggregations to registerAggregation
and remove their PROTOTYPEs.

Relates to #17085
2016-04-15 09:53:56 -04:00
Mike McCandless a361f5b810 fix test method name 2016-04-15 09:47:08 -04:00
Mike McCandless 529087301a fix IndexingMemoryController to use existing Settings API support to parse % or byte size, and to enforce min/max values; don't try to share string constants for settings in tests 2016-04-15 09:42:29 -04:00
Christoph Büscher 068df5f8aa Merge pull request #17756 from cbuescher/no-pfm-in-parser
Remove ParseFieldMatcher from AbstractXContentParser
2016-04-15 15:21:27 +02:00
Christoph Büscher de036d63d8 Rename context.parseFieldMatcher() to context.getParseFieldMatcher 2016-04-15 15:15:32 +02:00
Christoph Büscher 15c59d07b3 Remove ParseFieldMatcher from AbstractXContentParser
Currently we are able to set a ParseFieldMatcher on XContentParsers,
mainly to conveniently carry it around to be available where the
actual parsing happens. This was just recently introduced together
with ObjectParser so that ObjectParser can make use of deprecation
logging and throwing errors while parsing.

This however is trappy because we create parsers in so many places in
the code and it is easy to forget setting the right ParseFieldMatcher.
Instead we should hold the ParseFieldMatcher only in the parse contexts
(e.g. QueryParseContext).

This PR removes the ParseFieldMatcher from XContentParser. ObjectParser
can still make use of it because we can make the otherwise unbounded
`context` type to extend an interface that makes sure contexts used in
ObjectParser can supply a ParseFieldMatcher. Contexts in ObjectParser
are now no longer optional, but it is sufficient to pass in a small
lambda expression in places where no other context is available.

Relates to #17417
2016-04-15 15:14:46 +02:00
Nik Everett e3c65408ec Cut terms aggregation to registerAggregation
and remove its PROTOTYPE. This is the first aggregation builder that
serializes its targetValueType so ValuesSourceAggregatorBuilder had to
grow support for that.

Relates to #17085
2016-04-15 09:12:29 -04:00
Nik Everett cf80b00507 Cut filters aggregation to registerAggregation
and remove its PROTOTYPE.
2016-04-15 09:04:46 -04:00
Nik Everett 135511134c Fix bad toXContent for derivative aggregation
I busted it in the last commit.
2016-04-15 08:51:22 -04:00
Nik Everett a36b6138d7 Cut avg aggregation to registerAggregation
and remove its PROTOTYPE

Relates to #17085
2016-04-15 08:37:22 -04:00
Nik Everett b10c1f1939 Cut derivative aggregation to registerPipelineAggregation
and remove its PROTOTYPE.
2016-04-15 08:28:18 -04:00
Nik Everett beafae8ea5 Cut range aggregations to registerAggregation
and remove their PROTOTYPES. Does not remove the PROTOTYPEs from their
Range implementations which will have to wait for another commit.
2016-04-15 08:14:05 -04:00
Mike McCandless 36e99c5f8d switch to registered Settings for all IndexingMemoryController settings 2016-04-15 06:10:30 -04:00
Colin Goodheart-Smithe d863cbaa07 [TEST] fix geo_bounding_box tests to work with unmapped fields 2016-04-15 10:10:43 +01:00
Daniel Mitterdorfer 3688629e11 Adjust line-length of transport related classes to coding standard 2016-04-15 10:12:24 +02:00
Ryan Ernst c3f4eb36e3 Merge pull request #17768 from rjernst/dots4
Mappings: Fix array parsing to remove its context when finished parsing
2016-04-15 00:22:55 -07:00
LeonardGC 0b8be7f894 Update field-mapping.asciidoc (#17670) 2016-04-15 09:22:38 +02:00
Jason Tedor f985cfc95a Add max map count check
This commit adds a bootstrap check on Linux for the max map count (max
virtual memory areas) available to the Elasticsearch process.

Relates #16944
2016-04-14 21:04:03 -04:00
Igor Motov f030796b0b Shard level tasks in Bulk Action lose reference to their parent tasks
During the bulk action a hierachy of tasks is getting created: bulk->bulk[s] (coord node) -> bulk[s] (primary shard node) -> bulk[s][p] and bulk[s][r]. Due to a bug the first bulk[s] task didn't have bulk's task id is set as a parent id.  This commit fixes this bug.
2016-04-14 19:30:07 -04:00
Ryan Ernst 24130e1f30 Merge pull request #17764 from rjernst/parse_doc_without_threadlocal
Internal: Remove threadlocal from document parser
2016-04-14 15:17:45 -07:00
Ryan Ernst 6d8ab56350 Merge pull request #17769 from rjernst/gradle_idea_excludes
Build: Exclude gradle and eclipse build dirs from intellij
2016-04-14 15:16:05 -07:00
Ryan Ernst ba2abbf37c Build: Exclude gradle and eclipse build dirs from intellij
Intellij has a model of "everything is a source dir unless you say
otherwise". This change fixes the intellij configuration to not think
the gradle or eclipse build dirs are source dirs.
2016-04-14 14:54:52 -07:00
Ryan Ernst 60df3450f4 Mappings: Fix array parsing to remove its context when finished parsing
Handling of the current path when parsing a document is very sensitive.
This fixes a subtle bug in array parsing, where the path that was added
by parsing an array would not be cleared. It also adds a hard state
check at the end of parsing to ensure we ended with a clean path.
2016-04-14 14:45:29 -07:00
Ryan Ernst c6dd17a2c6 Internal: Remove threadlocal from document parser
The doc parser uses a context object to store the state of parsing,
namely the existing mappings, new mappings, and the parsed document.
Currently this uses a threadlocal which is "reset" for each doc parsed.
However, the thread local doesn't actually save anything, since
resetting is constructing new objects. This change removes the thread
local, which also simplifies the mapper service as it now does not need
to be closeable.
2016-04-14 13:10:20 -07:00
bloublou 83944c5628 Typo correction heap_size.asciidoc (#17745)
* Typo correction Xms Xmx

Typo correction on "-Xms4000mb -Xmx4000mb"

* Change mb to m for Xms/Xmx
2016-04-14 20:37:37 +02:00
Jason Tedor ea26e86880 Actually restore lost Node#start no-op check
This commit really reverts the inadvertent removal of allowing duplicate
calls to Node#start to be a no-op (but was mistakenly restored to
Node#stop in ddfa3a661510f25c2ce431dfd6fb86ac11eb8888).
2016-04-14 14:26:04 -04:00
Jason Tedor f87c599ff1 Revert "Restore lost Node#start no-op check"
This reverts commit ddfa3a661510f25c2ce431dfd6fb86ac11eb8888.
2016-04-14 14:26:04 -04:00
jaymode f35cfc3715 Copy checkstyle config to allow running from a jar
The checkstyle configuration files were being accessed as resources within the project and
being converted from a URL to a File by gradle. This works when the build tools project is being
referenced as a local project. However, when using the published jar the URL points to a resource
in the jar file, that URL cannot be converted to a File object and causes the build to fail.

This change copies the files into a `checkstyle` directory in the project build folder and always uses
File objects pointing to the copied files.
2016-04-14 13:33:10 -04:00
Nik Everett 3197f7f7d8 Remove PROTOTYPE from filter aggregation
and cut it to registerAggregation

Relates to #17085
2016-04-14 12:37:40 -04:00
Adrien Grand 0c9aca94d4 Disallow fielddata loading on text fields that are not indexed. #17747
If a field is not indexed, we cannot load fielddata for it.
2016-04-14 18:11:11 +02:00