52a70d8 | Rene Groeschke | 04 December 2020, 12:03:29 UTC | Update gradle to latest nightly | 04 December 2020, 12:03:29 UTC |
e48f038 | Rene Groeschke | 01 December 2020, 12:02:34 UTC | Disable toolchain autodetection for now | 04 December 2020, 11:36:38 UTC |
2b7f0a8 | Rene Groeschke | 01 December 2020, 11:22:17 UTC | Rework toolchain resolution filter | 04 December 2020, 11:36:38 UTC |
4f40d9b | Rene Groeschke | 30 November 2020, 21:15:49 UTC | Fix plugin func test setup | 04 December 2020, 11:36:38 UTC |
7816980 | Rene Groeschke | 30 November 2020, 21:08:35 UTC | Configure defaults for toolchain resolution in gradle func tests | 04 December 2020, 11:36:37 UTC |
dbaacf8 | Rene Groeschke | 30 November 2020, 20:26:58 UTC | Add debug output for resolved jdk toolchains | 04 December 2020, 11:36:37 UTC |
2c3a8f3 | Rene Groeschke | 30 November 2020, 18:16:25 UTC | Remove debug output | 04 December 2020, 11:36:37 UTC |
8531bc6 | Rene Groeschke | 30 November 2020, 18:10:45 UTC | Fix assertions for up-to-date Transforms - with gradle 6.8 artifact transforms output is kept in build cache - therefore there is no skipping info seen in our tests | 04 December 2020, 11:36:37 UTC |
ba3fd31 | Rene Groeschke | 30 November 2020, 12:32:08 UTC | Add RUNTIME_JAVA_HOME as resolvable tool chain | 04 December 2020, 11:36:36 UTC |
f86fc42 | Rene Groeschke | 30 November 2020, 12:25:08 UTC | Add debug output for resolving jdks | 04 December 2020, 11:36:36 UTC |
f9a9f68 | Rene Groeschke | 30 November 2020, 11:56:45 UTC | Update gradle to 6.8-rc-1 - resolve available gradle versions using built-in toolchain support - fixes deprecated usage of JavaInstallationRegistry | 04 December 2020, 11:36:36 UTC |
d071657 | David Roberts | 04 December 2020, 11:34:51 UTC | [ML] Remove unnecessary BWC constants (#65872) These are no longer needed after the backport of #65561 | 04 December 2020, 11:34:51 UTC |
11d7ad2 | Rene Groeschke | 04 December 2020, 11:34:17 UTC | Provide an extract only distribution configuration for bwc configurations (#63599) Provide a configuration that can be used for bwc setups that allows bwc setups to create an extracted bwc version es distribution without creating a tar configuration as done by the default assemble configuration A later PR will change the bwc setup configuration to rely on this extractedAssemble configuration when building distributions for bwc testing. One step closer to finish off #62115 | 04 December 2020, 11:34:17 UTC |
0a5596b | Dimitris Athanasiou | 04 December 2020, 10:46:00 UTC | [ML] Simplify reverting to snapshot while datafeed is reassigning (#65814) Previously, we were trying to compare that the job had gone past its current model snapshot in order to determine whether we should perform recovery. However, this is not necessary. We can instead simply always perform recovery if the opening job has a running datafeed task. This greatly simplifies the code. Relates #65630 Fixes #65710 | 04 December 2020, 10:46:00 UTC |
d17e714 | Tanguy Leroux | 04 December 2020, 10:26:14 UTC | Change index removal reason when IndicesService is stopping (#65816) When IndicesService is stopping on a data node, it closes every IndexService instances that are existing by calling the removeIndex(Index, IndexRemovalReason, String) method. The IndexRemovalReason that is passed as a parameter in this case is NO_LONGER_ASSIGNED which is also the one passed in other situation like removing an index because it got assigned to another data node. The fact that the same reason is used in multiple cases make it difficult for IndexEventListener to do the distinction between an index closed because the node is shutting down and an index closed between it moved away. This commit changes the IndexRemovalReason used when closing the IndicesService to be SHUTDOWN. | 04 December 2020, 10:26:14 UTC |
4e7adad | Martijn van Groningen | 04 December 2020, 09:36:46 UTC | Remove more usages of .watches system index. (#65835) Make use of the new query watches api to simply list watches or to delete all watches using the delete watch api. Relates to #62501 | 04 December 2020, 09:36:46 UTC |
cbd5d12 | Rory Hunter | 04 December 2020, 09:24:09 UTC | Add infrastructure for managing system indices (#65604) Part of #61656. Add the necessary support for automatically creating and updating system indices. This works by making it possible to create a system index descriptor with all the information needed to manage the mappings, settings and aliases. Follow-up work will opt existing indices into this framework. | 04 December 2020, 09:24:09 UTC |
201b25e | Yang Wang | 04 December 2020, 07:02:38 UTC | Fix xpack info and usage reports for operator privileges (#65867) This is a follow-up PR for #65256 to fix the xpack info and usage reports for operator privilegs. In summary, this PR ensures: * _xpack does not report operator privileges because it is categorised under security * _xpack/usage reports operator privileges status under the security section * _license/feature_usage reports last used time of operator privileges. It is up to the downstream to filter out this report if necessary. | 04 December 2020, 07:02:38 UTC |
7f61898 | Henning Andersen | 04 December 2020, 06:43:43 UTC | Extract DiskUsageIntegTestCase (#65540) Extracted DiskUsageIntegTestCase from DiskThresholdDeciderIT to allow other tests to easily test functionality relying on disk usage. Relates #65520 | 04 December 2020, 06:43:43 UTC |
7f249e4 | Julie Tibshirani | 04 December 2020, 01:57:23 UTC | Remove unused member MappedFieldType#indexAnalyzer. | 04 December 2020, 01:57:23 UTC |
aab655c | Julie Tibshirani | 04 December 2020, 00:07:06 UTC | Fix InternalEngineTests#testStoreHonorsLuceneVersion. This is a forward-port of a bugfix we found as part of backporting #65783. | 04 December 2020, 00:18:46 UTC |
9612c9a | Armin Braun | 03 December 2020, 23:40:23 UTC | Add Netty ByteBuf Leak Check to REST Test Clusters (#64528) We do this check in all tests that inherit from `EsTestCase` but didn't check for `ByteBuf` leaks in rest test clusters, which means we have very little coverage of the REST layer. With recent reports of very rare leak warnings in logs I think it's worthwhile to do this check in REST tests as well. | 03 December 2020, 23:40:23 UTC |
ac18f24 | Julie Tibshirani | 03 December 2020, 22:02:42 UTC | Avoid passing Lucene version to Store#createEmpty. (#65783) Store already holds onto index settings, so it can look up the version directly. This helps consolidate version checks in a smaller number of places. | 03 December 2020, 22:02:42 UTC |
5034b11 | Jay Modi | 03 December 2020, 21:49:27 UTC | Mute HistoryTemplateIndexActionMappingsTests (#65860) HistoryTemplateIndexActionMappingsTests testIndexActionFields has been failing in CI. This commit mutes the test. Relates #65091 | 03 December 2020, 21:49:27 UTC |
2ab9ad0 | Jay Modi | 03 December 2020, 21:31:44 UTC | Mute IndicesClientIT.testDataStreams (#65856) This commit mutes IndicesClientIT.testDataStreams as this test is failing in CI intermittently. Relates #60746 Relates #60461 | 03 December 2020, 21:31:44 UTC |
a44ad56 | Jack Conradson | 03 December 2020, 20:01:07 UTC | Complete replacing member data with decorations in the ir tree (#64825) This change replaces all the member data in the ir nodes with decorations instead. This completes the transition to a decoration system in the ir tree. This change allows for maximum flexibility when modifying existing phases or adding additional phases. | 03 December 2020, 20:01:07 UTC |
56a25bf | Seth Michael Larson | 03 December 2020, 19:47:13 UTC | Mark Task APIs as experimental in rest-api-spec | 03 December 2020, 19:47:13 UTC |
46584db | Thlamz | 03 December 2020, 19:30:06 UTC | Pass aggregator suppliers as aggregator factory constructor arguments (#65466) Preventing a double call to the ValuesSourceRegistry by saving the return value of the first call on ValuesSourceAggregationBuilder#doBuild and passing it onto innerBuild which then gets picked up by the Aggregator's builder and saved in the Aggregator's factory to be used in its doCreateInternal method. | 03 December 2020, 19:30:06 UTC |
2c33f9f | Gordon Brown | 03 December 2020, 19:18:00 UTC | Add basic Fleet plugin (#65275) A very basic plugin to manage Fleet system indices. Currently, just registers these patterns as system indices: - `.fleet-servers*` - `.fleet-policies*` - `.fleet-agents*` - `.fleet-actions*` | 03 December 2020, 19:18:00 UTC |
48add56 | Jay Modi | 03 December 2020, 18:35:22 UTC | Mute TimeThrottleIntegrationTests.testTimeThrottle (#65844) This test is failing on master in CI occasionally. This change mutes the test to stop the noise until it can be looked at again. Relates #65176 | 03 December 2020, 18:35:22 UTC |
638402c | Bogdan Pintea | 03 December 2020, 18:19:15 UTC | Abort sorting in case of local agg sort queue overflow (#65687) In case the local agg sorter queue gets full and no limit has been provided, the local sorter will now erroneously call the failure callback for every single row in the original rowset that's left over the local queue limit (instead for just the first one). The failure response is dispatched in any case, so this is relatively harmless. The sorter continues iterating on the original response fetching subsequent pages. In case of correct Elasticsearch behaviour, this is also harmless, it'll just trigger a number of internal exceptions. However, in case of a pagination defect in Elasticsearch (like GH#65685, where the same search_after is returned), this will result in an effective spin loop, potentially rendering eventually the node unresponsive. This PR simply breaks both the inner loop iterating over the current unsorted rowset, as well as the outer one, iterating over the left pages. It also fixes an outdated documentation limitation. | 03 December 2020, 18:19:15 UTC |
491ab69 | Tom Veasey | 03 December 2020, 17:59:05 UTC | [ML] Improve robustness of test to perturbations in the ml-cpp results (#65839) This test should check the count of low probability results. This is sufficient to test the rule condition is applied and is more stable to perturbations in results. | 03 December 2020, 17:59:05 UTC |
49e492f | David Roberts | 03 December 2020, 17:18:08 UTC | [ML] Adding assignment_memory_basis to model_size_stats (#65561) At present the Java code makes a decision on whether to use current model memory or model memory limit to calculate how much memory a job requires to be assigned. The plan is to move this decision to the C++ code, which will report it via a new field in the model size stats. An additional change will be that once we have made the switch from using model memory limit to using current model memory we will never switch back, as this causes large fluctuations up and down in memory requirement which will be much more noticeable when autoscaling is in use. Although the only two options at present are model memory limit and current model memory, the new enum includes a third possibility, peak model memory. To switch to this now would be tricky, as there have been two bugs in the implementation of peak model memory which render its value unreliable in 7.x. However, in 8.x it might make sense to switch to using peak model memory instead of current model memory and it's much easier from a BWC perspective if the enum contains all the values from the start. Relates #63163 | 03 December 2020, 17:18:08 UTC |
2044caa | James Rodewig | 03 December 2020, 17:14:38 UTC | [DOCS] EQL: Document ? wildcard (#65698) | 03 December 2020, 17:14:38 UTC |
fd3d7e3 | Alexander Reelsen | 03 December 2020, 16:55:18 UTC | Remove class that is part of commons-codec (#65259) This class was copied from the trunk and is now part of the stable release since 2012. | 03 December 2020, 16:55:18 UTC |
12c9ee4 | Mike Barretta | 03 December 2020, 16:48:45 UTC | Update inference-bucket-aggregation.asciidoc tiny change to properly align the first code example and to add a missing word | 03 December 2020, 16:48:45 UTC |
cb99487 | Przemko Robakowski | 03 December 2020, 16:01:15 UTC | Add hidden data streams to stats endpoint (#65795) This change adds support for hidden data streams in stats endpoint with usual _data_stream/<index>/_stats?expand_wildcards=hidden syntax. Closes #65323 | 03 December 2020, 16:01:15 UTC |
4d084c9 | Dan Hermann | 03 December 2020, 15:12:28 UTC | Add missing comma in sample payload for the watcher's pagerduty action (#65819) | 03 December 2020, 15:12:28 UTC |
8c29e54 | Christoph Büscher | 03 December 2020, 15:10:55 UTC | Adapt test skip versions after backport | 03 December 2020, 15:10:55 UTC |
0693f01 | Benjamin Trent | 03 December 2020, 14:57:06 UTC | [ML] muting snapshot upgrader bwc test (#65824) | 03 December 2020, 14:57:06 UTC |
e955f77 | James Rodewig | 03 December 2020, 14:55:47 UTC | [DOCS] Fix typo in histogram agg docs (#65822) | 03 December 2020, 14:55:47 UTC |
f0a58c9 | Ioannis Kakavas | 03 December 2020, 14:53:20 UTC | Apply fips.gradle to the correct plugins (#65807) Currently the sections wrapped with pluginManager.withPlugin('java') were not applied to all places they ought to. | 03 December 2020, 14:53:20 UTC |
b554d5b | Christos Soulios | 03 December 2020, 14:40:09 UTC | Store _doc_count field as custom term frequency (#65776) A while back, Lucene introduced the ability to index custom term frequencies, ie. giving users the ability to provide a numeric value that should be indexed as a term frequency rather than letting Lucene compute the term frequency by itself based on the number of occurrences of a term. This PR modifies the _doc_count field so that it is stored as Lucene custom term frequency. A benefit of moving to custom term frequencies is that Lucene will automatically compute global term statistics like totalTermFreq which will let us know the sum of the values of the _doc_count field across an entire shard. This could in-turn be useful to generalize optimizations to rollup indices, e.g. buckets aggregations where all documents fall into the same bucket. Relates to #64503 | 03 December 2020, 14:40:09 UTC |
dd50520 | Benjamin Trent | 03 December 2020, 14:16:06 UTC | [ML] Fix snapshot upgrader so that if state is not fully written or parseable the task fails (#65755) It is possible that snapshot upgrader execution path continues before the old model state is fully read by the native process. To prevent this, a flush request is made after the state is loaded. This is to verify that the all the state has been read by the native process. This allows the task to fail if reading the state fails and prevents some strange race conditions. closes https://github.com/elastic/elasticsearch/issues/65699 | 03 December 2020, 14:16:06 UTC |
ccad78e | Benjamin Trent | 03 December 2020, 13:40:31 UTC | [ML] make InferenceIngestIT more lenient when checking cache miss counts (#65774) Looking over the failure history, it is always the cache miss count that is off. This is mostly ok as all the failures had indicated that there were indeed cache failures and every one of them were fence-post errors. Opting to make the cache miss count check lenient as other stats checked verify consistency. closes https://github.com/elastic/elasticsearch/issues/61564 | 03 December 2020, 13:40:31 UTC |
3adb6d8 | Martijn van Groningen | 03 December 2020, 13:14:55 UTC | Add query watches api to retrieve multiple watches. (#64582) This api supports pagination (from / size) and querying and sorting by watch _id and watcher metadata. This avoids using .watch index directly. On a per watch basis the same information that the get watch api returns is returned, except version. Relates #62501 | 03 December 2020, 13:14:55 UTC |
1a3fefb | David Turner | 03 December 2020, 10:55:53 UTC | Clarify usage of jvm.options in its comments (#61456) Since #51882 we recommend not editing the `jvm.options` file, preferring instead to override its contents with additional files in `jvm.options.d`. However the inline comments in this file do not point users in that direction. This commit adjusts these inline comments. | 03 December 2020, 10:55:53 UTC |
47cb23c | David Turner | 03 December 2020, 10:16:19 UTC | Explain size limit for transport messages (#65803) If a node receives an unreasonably large transport message then it rejects it and closes the connection. This can occur if, for instance, a rather small node tries to join a cluster with a rather oversized cluster state. The message logged in this case says that the message is too large but does not give any clues to the user what to do about it. This commit adds to the message the information that the limit is 30% of the node's heap size, to give a hint that a larger heap might help. | 03 December 2020, 10:16:19 UTC |
f270a05 | Howard | 03 December 2020, 10:14:19 UTC | Remove immediately reversed Nested ID FieldMapper configuration (#65763) The nested id FieldType in IdFieldMapper sets its stored configuration to `true` and then immediately sets it to `false`. | 03 December 2020, 10:14:19 UTC |
1c8a977 | Christos Soulios | 03 December 2020, 10:08:00 UTC | Lower minimum compatibility version of _doc_count field tests (#65790) After merging _doc_count field type in v7.11.0 (#64594), this PR lowers the minimum compatibility version from v8.0.0 to v7.11.0 Relates to #64503 | 03 December 2020, 10:08:00 UTC |
1ff8baa | Albert Zaharovits | 03 December 2020, 08:47:41 UTC | Misc integ tests for DLS and FLS (#65801) Misc integ tests for security DLS and FLS for terms lookup, geo shape and percolate queries. Co-authored-by: Yang Wang yang.wang@elastic.co | 03 December 2020, 08:47:41 UTC |
def2f27 | Yang Wang | 03 December 2020, 05:49:33 UTC | Phase 1 support for operator privileges (#65256) In some Elastic Stack environments, there is a distinction between the operator of the cluster infrastructure and the administrator of the cluster. This distinction cannot be supported currently because the "administrator" often has the superuser role which grants each and every privilege of the cluster. This PR adds a new feature to protect a fixed set of APIs from the "administrator" even when it is a highly privileged user such as superuser. It enhances the Elasticsearch security model to have an additional layer of restriction in addition to the RBAC. Co-authored-by: Tim Vernum <tim@adjective.org> | 03 December 2020, 05:49:33 UTC |
7518b1c | Nhat Nguyen | 03 December 2020, 01:47:47 UTC | Do not reuse test cluster in CrossClusterSearchIT The `testProxyConnectionDisconnect` test can use a node without the remote_cluster_client that is created in `testRemoteClusterClientRole`. | 03 December 2020, 01:48:59 UTC |
26350df | Nhat Nguyen | 02 December 2020, 22:41:33 UTC | Ensure notify when proxy connections disconnect (#65697) TransportService doesn't respond to the pending requests of proxy connections when the underlying connections get disconnected because proxy connections do not override the getCacheKey method. Some CCS requests would never be completed because of this bug. | 02 December 2020, 22:41:33 UTC |
667908e | Armin Braun | 02 December 2020, 20:58:55 UTC | Add WARN Logging for Slow REST Request Handling (#65748) Same as #62444 but for REST requests. | 02 December 2020, 20:58:55 UTC |
68358df | Benjamin Trent | 02 December 2020, 20:00:58 UTC | [ML] adding ml autoscaling integration test (#65638) This adds ml autoscaling integration tests. The test verifies that the scaling requirements adjust according to the current real load on the cluster given machine learning jobs of various sizes. Additionally, there was a bug in the ml scaling service settings. This commit addresses the bug. | 02 December 2020, 20:00:58 UTC |
ff5cb90 | David Turner | 02 December 2020, 19:30:55 UTC | Remove escape hatch permitting incompatible builds (#65753) Today in `7.x` there is a deprecated system property that bypasses the check that prevents nodes of incompatible builds from communicating. This commit removes the system property in `master` so that the check is always enforced. Relates #65601, #65249 | 02 December 2020, 19:30:55 UTC |
e4e8905 | Tal Levy | 02 December 2020, 17:33:38 UTC | Fixes GeoLineAggregator bugs (#65521) This commit unmutes and fixes tests around some GeoLineAggregator edge cases. - MergedGeoLines had a silly bug where it was accepting InternalGeoLines that were empty - "complete" is measured by the heap-mode of the BucketedSort, which is a problem since if the length of the data equals the max-size, then it is difficult to know whether any values were discarded. - GeoLineBucketSort had an array-resizing bug s/>/>= Fixes #65473. | 02 December 2020, 17:33:38 UTC |
ae69472 | Bhavya Gupta | 02 December 2020, 17:23:12 UTC | Updated painless-walkthrough documentation (#65530) Update documentation about regex note. | 02 December 2020, 17:28:42 UTC |
8dc71c3 | Mark Vieira | 02 December 2020, 17:27:24 UTC | Always use "elasticsearch" as Gradle project name (#65709) | 02 December 2020, 17:27:24 UTC |
a1b6be0 | James Rodewig | 02 December 2020, 17:09:53 UTC | [DOCS] Fix docs integ tests for release builds (#65761) | 02 December 2020, 17:09:53 UTC |
3b859f5 | Przemyslaw Gomulka | 02 December 2020, 16:36:31 UTC | Compress audit logs (#64472) audit logs should be compressed when rolling over due to size based triggering policy breaching 1GB. Files are not being deleted. closes #63843 | 02 December 2020, 16:36:31 UTC |
894dbc7 | Hendrik Muhs | 02 December 2020, 16:00:19 UTC | [Transform] fix intermittent test_usage failure (#65742) relax test for trigger count, it might have been incremented but not persisted fixes #52931 | 02 December 2020, 16:00:19 UTC |
95fda0a | Hendrik Muhs | 02 December 2020, 15:58:52 UTC | [Transform] Use test threadpool to avoid conflict with other tests (#65721) use a test threadpool instead of the generic threadpool in test fixes #65542 | 02 December 2020, 15:58:52 UTC |
0f406f1 | James Rodewig | 02 December 2020, 15:37:01 UTC | [DOCS] Add cluster get settings API example (#65754) | 02 December 2020, 15:37:01 UTC |
ab08631 | David Kyle | 02 December 2020, 15:20:53 UTC | Add search runtime_mappings to datafeed configuration (#65606) Adds the runtime_mappings section to DatafeedConfig to define runtime fields. The runtime fields are then passed through to datafeed searches | 02 December 2020, 15:20:53 UTC |
8e3cc30 | James Rodewig | 02 December 2020, 14:46:02 UTC | [DOCS] Remove inert component template file (#65749) | 02 December 2020, 14:46:02 UTC |
7262187 | James Rodewig | 02 December 2020, 14:42:35 UTC | [DOCS] Remove erroneous `flat_settings` query param (#65670) (#65745) Co-authored-by: Thiago Souza <thiago@elastic.co> | 02 December 2020, 14:42:35 UTC |
20c82c7 | Dimitris Athanasiou | 02 December 2020, 14:29:00 UTC | [ML] Mute failing MlDistributedFailureIT test (#65747) Relates #64710 | 02 December 2020, 14:29:00 UTC |
483bed8 | David Turner | 02 December 2020, 14:21:18 UTC | Fix up BWC after backport of #65732 as #65601 (#65741) Relates #65732 Relates #65601 | 02 December 2020, 14:21:18 UTC |
bcea87f | Howard | 02 December 2020, 13:51:39 UTC | [DOCS] Fix EQL syntax formatting (#65711) | 02 December 2020, 13:51:39 UTC |
90d25d6 | Dan Hermann | 02 December 2020, 13:22:40 UTC | [DOCS] URI parts processor (#65695) | 02 December 2020, 13:22:40 UTC |
64a1425 | David Turner | 02 December 2020, 13:18:28 UTC | Disable BWC for backport of #65732 as #65601 | 02 December 2020, 13:18:28 UTC |
6f11752 | James Rodewig | 02 December 2020, 13:13:43 UTC | [DOCS] Fix `_doc_count` field title (#65704) | 02 December 2020, 13:13:43 UTC |
aba2f3e | David Turner | 02 December 2020, 13:08:23 UTC | Validate build hash in handshake (#65732) There is no guarantee of wire compatibility between nodes running different builds of the same version, but today we do not validate whether two communicating nodes are compatible or not. This results in confusing failures that look like serialization bugs, and it usually takes nontrivial effort to determine that the failure is in fact due to the user running incompatible builds. This commit adds the build hash to the transport service handshake and validates that matching versions have matching build hashes. Closes #65249 | 02 December 2020, 13:08:23 UTC |
6f323ad | István Zoltán Szabó | 02 December 2020, 13:07:17 UTC | [DOCS] Changes wording of pivot parameter in PUT transforms API docs. (#65731) | 02 December 2020, 13:07:17 UTC |
01c343f | David Turner | 02 December 2020, 11:59:01 UTC | Fix BWC after backport of #65564 (#65727) Relates #65564 Relates #65689 | 02 December 2020, 11:59:01 UTC |
e066c0c | Rory Hunter | 02 December 2020, 11:28:24 UTC | Try to make DeprecationHttpIT more robust (#65665) Closes #65589 hopefully. There is a test case that checks whether deprecation logs have been indexed. It uses `assertBusy` so that it keeps checking for a period, since it may take some time for the data to become available. However, the test nonetheless occasionally still fails, with no shards being available to search. In an attempt to address this, add a `_refresh` call on the deprecation data stream, and increase the `assertBusy` timeout to 30s. | 02 December 2020, 11:28:24 UTC |
2ebab21 | David Turner | 02 December 2020, 10:59:15 UTC | Disable BWC tests for backport of #65689 | 02 December 2020, 10:59:15 UTC |
b2b400b | David Turner | 02 December 2020, 10:47:52 UTC | Nix bogus assertion in getTimestampMillisRange() (#65720) This method is called without any guarantee that the shard hasn't been closed, so asserting that the shard is active is bogus. Instead we can proceed no matter what state the shard is in, and let the engine throw an `AlreadyClosedException` if needed. Closes #65713 | 02 December 2020, 10:47:52 UTC |
f0fc1b3 | Alan Woodward | 02 December 2020, 10:06:40 UTC | Use scriptless fields in CoreTestTranslator (#65599) CoreTestTranslator re-writes some core search yaml tests into a tests for runtime queries, and mimics source-only fields by building ad-hoc painless scripts for each runtime type. We now have source-only fields built in via scriptless mappings, so we can cut over to using these instead. | 02 December 2020, 10:06:40 UTC |
2adac36 | Jim Ferenczi | 02 December 2020, 07:54:40 UTC | Create async search index if necessary on updates and deletes (#64606) This change ensures that we create the async search index with the right mappings and settings when updating or deleting a document. Users can delete the async search index at any time so we have to re-create it internally if necessary before applying any new operation. | 02 December 2020, 07:54:40 UTC |
19dfa7b | Bogdan Pintea | 01 December 2020, 21:15:53 UTC | Tableau connector: read version to build from src (#65671) This changes the Tableau connector building script to be able to read the version of the connector to be built straight from Version.java file, if that's available (which is the case, now being in tree). A new parameter has been added, to allow specifying a version qualifier. Also, the 'alias' parameter is made optional. | 01 December 2020, 21:15:53 UTC |
3c3a432 | Christoph Büscher | 01 December 2020, 20:40:27 UTC | Support unmapped fields in search 'fields' option (#65386) Currently, the 'fields' option only supports fetching mapped fields. Since 'fields' is meant to be the central place to retrieve document content, it should allow for loading unmapped values. This change adds implementation and tests for this feature. Closes #63690 | 01 December 2020, 20:40:27 UTC |
b4233f0 | Gil Raphaelli | 01 December 2020, 18:53:33 UTC | [DOCS] Fix _doc_count example typo (#65686) | 01 December 2020, 18:53:33 UTC |
8d5f101 | Benjamin Trent | 01 December 2020, 18:32:48 UTC | [ML] test mute for testUpgradeJobSnapshot (#65700) relates to https://github.com/elastic/elasticsearch/issues/65699 | 01 December 2020, 18:32:48 UTC |
1c3ddf8 | James Rodewig | 01 December 2020, 17:56:12 UTC | [DOCS] EQL: Flatten EQL syntax headings (#65693) | 01 December 2020, 17:56:12 UTC |
c327794 | Christoph Büscher | 01 December 2020, 17:49:50 UTC | Fix range query on date fields for number inputs (#63692) Currently, if you write a date range query with numeric 'to' or 'from' bounds, they can be interpreted as years if no format is provided. We use "strict_date_optional_time||epoch_millis" in this case that can interpret inputs like 1000 as the year 1000 for example. This PR change this to always interpret and parse numbers with the "epoch_millis" parser if no other formatter was provided. Closes #63680 | 01 December 2020, 17:49:50 UTC |
dc64498 | James Rodewig | 01 December 2020, 16:38:40 UTC | [DOCS] Fix URL in rollup API JSON spec (#65683) | 01 December 2020, 16:38:40 UTC |
ac1dbb7 | James Rodewig | 01 December 2020, 16:30:17 UTC | [DOCS] EQL: Remove outdated wildcard ref (#65684) | 01 December 2020, 16:30:17 UTC |
08df709 | Dimitris Athanasiou | 01 December 2020, 16:25:03 UTC | [ML] Update serialization versions for revert snapshot force param (#65680) | 01 December 2020, 16:25:03 UTC |
44a60d0 | Rene Groeschke | 01 December 2020, 16:23:53 UTC | Add 6.8 branch to packer cache script (#65510) * Add 6.8 branch to packer cache script * use --reference on cloning 6.8 brach to speed cloning up a bit * Remove 6.8 branch checkout after resolving deps to keep image small | 01 December 2020, 16:23:53 UTC |
b902d80 | David Turner | 01 December 2020, 15:41:48 UTC | Record timestamp field range in index metadata (#65564) Queries including a filter by timestamp range are common in time-series data. Moreover older time-series indices are typically made read-only so that the timestamp range becomes immutable. By recording in the index metadata the range of timestamps covered by each index we can very efficiently skip shards on the coordinating node, even if those shards are not assigned. This commit computes the timestamp range of immutable indices and records it in the index metadata as the shards start for the first time. Note that the only indices it considers immutable today are ones using the `ReadOnlyEngine`, which includes frozen indices and searchable snapshots but not regular indices with a write block. | 01 December 2020, 15:41:48 UTC |
9ff95eb | Dimitris Athanasiou | 01 December 2020, 14:16:41 UTC | [ML] Add unit tests for DatafeedContext and RestartTimeInfo (#65674) These were meant to go with #65630 but were left out. | 01 December 2020, 14:16:41 UTC |
318340a | Dimitris Athanasiou | 01 December 2020, 14:16:30 UTC | [ML] Reduce log noise in BaseMlIntegTestCase (#65675) Tests inheriting `BaseMlIntegTestCase` have noisy logs with the error `No processor type exists with name [...]`. Those are caused because of loaded templates that need the `IngestCommonPlugin` loaded. This commit adds a test dependency to the `ingest-common` module and loads `IngestCommonPlugin` in `BaseMlIntegTestCase` to reduce such noise. | 01 December 2020, 14:16:30 UTC |
eac2104 | James Rodewig | 01 December 2020, 14:00:45 UTC | [DOCS] Label legacy rollup APIs (#65518) | 01 December 2020, 14:00:45 UTC |
581e5c8 | James Rodewig | 01 December 2020, 13:57:07 UTC | [DOCS] Update rollup glossary item (#65519) Co-authored-by: Lisa Cawley <lcawley@elastic.co> | 01 December 2020, 13:57:07 UTC |
6b0a1df | Benjamin Trent | 01 December 2020, 12:51:36 UTC | [ML] fixes minor snapshot upgrader task failure bug and rest client test failure (#65607) This commit fixes two issues: - On task failure and when the task has completed, there is no reason to utilize the single queued executor. - The high level rest client tests should "await_completion" to make sure the task has completed before allowing other tests to continue. closes https://github.com/elastic/elasticsearch/issues/65364 | 01 December 2020, 12:51:36 UTC |
83488ff | Benjamin Trent | 01 December 2020, 12:49:16 UTC | [ML] update ChunkedTrainedModelPersisterTests to signal latch on persistence Minor test bug ChunkedTrainedModelPersisterTests was not mocking the inference index refresh call. Consequently the latch was not being signaled. | 01 December 2020, 12:49:16 UTC |
c0ff8ec | Dimitris Athanasiou | 01 December 2020, 12:35:09 UTC | [ML] Resume reassigning datafeed from loaded model snapshot (#65630) This fixes a long outstanding issue with the resume behaviour of a datafeed that has been reassigned on a node. When we resume a datafeed, we start from the latest result time or the latest record time, whichever is greater. This makes sense when a job had been closed gracefully previously. In the scenario when a job is being reassigned, it is highly likely that the job has seen data after the latest snapshot was persisted. This means that with the current behaviour, we load that snapshot and then we resume sending data from the point the job had reached previously. This results to the model having a knowledge gap. This commit addresses this by checking whether a reassining datafeed has gone past its job's model snapshot and reverting the job to that snapshot while deleting intervening results. Then we resume the datafeed from the latest result or latest record of the snapshot. Fixes #63400 | 01 December 2020, 12:35:09 UTC |