| b174af6 | Rene Groeschke | 20 April 2022, 10:08:52 UTC | Apply precommit checks for non productive projects (#85533) (#86027) Ensure projects with only yaml, java or cluster tests also apply precommit checks. We only apply testingconventions now for projects with existing src test folder as the TestingConventionsTask is incompatible with projects with no test sourceSet. This is a prerequisite to port more projects away from using StandaloneRestTestPlugin and RestTestPlugin in favor of yaml, java or cluster tests with dedicated sourceSets. Also we fix deprecation warnings for forbiddenPattern and filePermissions tasks about implicit non declared dependencies on resources tasks | 20 April 2022, 10:08:52 UTC |
| a8be2fa | Yannick Welsch | 20 April 2022, 07:59:07 UTC | Rewrite match_all inside must_not (#85999) (#86028) A must_not with a match_all clause inside a bool query is currently not rewritten to a match_none query. This means that running a boolean query with "must_not":[{"terms":{"_tier":["data_frozen","data_cold"]}] is currently not rewritten as match_none on a cold/frozen tier node. | 20 April 2022, 07:59:07 UTC |
| aa6b946 | Lisa Cawley | 19 April 2022, 16:17:36 UTC | [DOCS] Add tagged section in breaking changes (#85965) | 19 April 2022, 16:17:36 UTC |
| 71c18aa | Artem Prigoda | 19 April 2022, 15:58:54 UTC | [DOCS] [8.2] Add 8.1.3 release notes (#86001) * [DOCS] Add 8.1.3 release notes * Update migrate_8_1.asciidoc | 19 April 2022, 15:58:54 UTC |
| a898db3 | Luca Cavanna | 19 April 2022, 14:33:32 UTC | [DOCS] add composite field context to the painless execute docs (#85513) (#86000) The painless execute API supports the composite_field context since the composite field was added. This commit adds docs for it where missing. Relates to #78050 | 19 April 2022, 14:33:32 UTC |
| 4b67752 | Adam Locke | 19 April 2022, 14:30:19 UTC | [DOCS] Remove conditional tagging for user profiles (#85998) Removes conditional tagging for user profiles in 8.2. | 19 April 2022, 14:30:19 UTC |
| 39c9414 | Nikola Grcevski | 19 April 2022, 14:02:51 UTC | Remove references to bootstrap.system_call_filter (#85964) (#85972) | 19 April 2022, 14:02:51 UTC |
| f9d3638 | Dario Gieselaar | 19 April 2022, 09:09:35 UTC | [8.2] Grant kibana_system role read access to APM data streams (#85744) (#85923) Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com> | 19 April 2022, 09:09:35 UTC |
| 01d3bf2 | David Turner | 19 April 2022, 07:10:58 UTC | More docs re. removing cluster.initial_master_nodes (#85948) (#85979) Ensures that on every page of the docs that mentions `cluster.initial_master_nodes` also mentions that this setting must be removed after bootstrapping completes. | 19 April 2022, 07:10:58 UTC |
| 2b4963c | Julie Tibshirani | 18 April 2022, 18:58:29 UTC | Remove custom KnnVectorFieldExistsQuery (#85945) This query was added to Lucene in version 9.1, so we can delete our version in favor of using the Lucene one. | 18 April 2022, 20:50:15 UTC |
| bf04816 | Mark Vieira | 18 April 2022, 19:36:16 UTC | Remove MacOS from platform support testing matrix MacOS is not an officially supported production platform. Maintaining this test infrastructure is costly and has historically been very unstable. While we won't maintain continuous integration here, we'll still continue to build and publish MacOS artifacts and support it in developoment environments. | 18 April 2022, 19:46:46 UTC |
| b847e5d | Ryan Ernst | 18 April 2022, 19:02:05 UTC | Relax data path deprecations from critical to warn (#85952) (#85963) The deprecations for index and shared data paths were defined at critical level, but these are not removed in 8.0. This commit relaxes these deprecation messages so they are just warnings. relates #85695 | 18 April 2022, 19:02:05 UTC |
| a1935d2 | Adam Locke | 18 April 2022, 13:26:41 UTC | [DOCS] Add detached mode for enrolling Kibana (#85933) (#85937) * [DOCS] Add detached mode for enrolling Kibana * Clarify commands for starting and enrolling Kibana Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com> | 18 April 2022, 13:26:41 UTC |
| cbd58a1 | Rory Hunter | 18 April 2022, 09:27:42 UTC | Adjust SQL expended test output Closes #85949. The employee test data has dates in it, and as time has moved on, the test data that matches a test SQL query has changed. Update the expected rows in the test. | 18 April 2022, 09:31:14 UTC |
| 098b120 | Gordon Brown | 15 April 2022, 22:18:37 UTC | Update license override list (#85940) (#85943) This commit updates the list of license expiration overrides as requested by our compliance team. | 15 April 2022, 22:18:37 UTC |
| ccfb350 | Mark Vieira | 14 April 2022, 21:16:54 UTC | Upgrade bytebuddy to latest version with Java 19 support (#85881) (#85916) | 14 April 2022, 21:16:54 UTC |
| d968fae | Julie Tibshirani | 14 April 2022, 16:56:37 UTC | Remove out-of-date note about kNN with filters We implemented this in #84734 but forgot to update these docs. | 14 April 2022, 17:18:35 UTC |
| 246131a | Yang Wang | 14 April 2022, 15:33:41 UTC | User Profile: Initial beta overview documentations (#83766) (#85907) Add initial feature overview documentation pages for User Profile and Security Domain. Co-authored-by: Adam Locke <adam.locke@elastic.co> Co-authored-by: Tim Vernum <tim@adjective.org> Co-authored-by: Albert Zaharovits <albert.zaharovits@gmail.com> Co-authored-by: Adam Locke <adam.locke@elastic.co> Co-authored-by: Tim Vernum <tim@adjective.org> Co-authored-by: Albert Zaharovits <albert.zaharovits@gmail.com> | 14 April 2022, 15:33:41 UTC |
| 88c08be | Salvatore Campagna | 14 April 2022, 14:07:54 UTC | Update docs for v8.2.0 BC3 (#85892) | 14 April 2022, 14:07:54 UTC |
| 2730fb6 | Adam Locke | 14 April 2022, 13:31:00 UTC | [DOCS] Add manage_enrich role to built-in roles (#85877) (#85900) | 14 April 2022, 13:31:00 UTC |
| eca5f49 | Yang Wang | 14 April 2022, 01:43:28 UTC | User Profile - Rename access to labels for request, response and mappings (#85797) (#85883) It is agreed that the "access" field should be renamed to the more generic "labels". Data stored in this field are meant to be application specific (similar to "data" but searchable) and ES does not attempt to understand its content or purpose. Hence the more generic name is a better fit. | 14 April 2022, 01:43:28 UTC |
| 7092b69 | Yang Wang | 13 April 2022, 23:47:36 UTC | User Profile - Remove profile.user.active field (#85856) (#85884) The profile document has two boolean fields, profile.enabled and profile.user.active. The intention was that profile.enabled=false means the profile is not searchable (somewhat equivalent to being deleted) and profile.user.active=false means the profile is searchable but may not be assignable (in Kibana). However we don't currently have a concrete requirement for the later. Also we don't provide an API so far to set profile.user.active to false. Hence we agree to leave this field out till we have a better requirement for it. | 13 April 2022, 23:47:36 UTC |
| 1afcc82 | Mark Vieira | 13 April 2022, 21:12:24 UTC | Add environment variables for Java 18 and 19 Java home | 13 April 2022, 21:13:32 UTC |
| 3c5d6c1 | Rory Hunter | 13 April 2022, 17:45:14 UTC | Upgrade to JDK 18.0.0+36 (#85376) (#85875) Closes #85357. Upgrade to the latest Adoptium JDK for bundling with Elasticsearch. | 13 April 2022, 17:45:14 UTC |
| b46ba91 | David Kyle | 13 April 2022, 13:25:14 UTC | [ML] Fix serialisation of text embedding updates (#85863) (#85868) | 13 April 2022, 13:25:14 UTC |
| 6ca9a15 | Przemyslaw Gomulka | 13 April 2022, 10:31:10 UTC | Prevent ThreadContext header leak when sending response backport(#68649) (#85860) We need to stash thread context in DefaultRestChannel before we call channel.sendResponse because when calling this method, our execution might be delayed and the thread be reused for another task - like sending another response. And it would see thread context from the initial “delayed” work. This commit also expands the assertions on the empty thread context to make sure it does not contains response headers. closes #68278 | 13 April 2022, 10:31:10 UTC |
| 00d0894 | Rory Hunter | 13 April 2022, 08:24:15 UTC | Correct docs/changelog/85581.yaml | 13 April 2022, 08:24:15 UTC |
| 4af5a2f | David Turner | 13 April 2022, 08:04:58 UTC | Fix TranslogTests#testStats (#85828) (#85857) Today these tests assume that the last-modified time on the translog file is at least one millisecond before the time at which stats are computed, but this isn't always true. With this commit we add a (bounded) wait for the clock to advance beyond the file's last-modified time to ensure that we observe a nonzero age. Closes #85717 | 13 April 2022, 08:04:58 UTC |
| 60d8398 | David Turner | 13 April 2022, 08:02:18 UTC | Reduce port range re-use in tests (#85777) (#85858) Today we select the port range to use in tests by taking the Gradle worker ID mod 223. We now use significantly more than 223 Gradle workers in each test run which means port ranges are re-used, resulting in spurious failures. This commit replaces the magic number 223 with a computation defined in terms of the available ports and the number of ports to allocate per worker. It also reduces the number of ports per worker to 30 to try and reduce the frequency of port clashes. | 13 April 2022, 08:02:18 UTC |
| 8ba1a5c | Artem Prigoda | 12 April 2022, 18:49:48 UTC | Expose proxy settings for GCS repositories (#85785) (#85833) They were added in #82737, but not exposed via the GCSPlugin to end users. | 12 April 2022, 18:49:48 UTC |
| 2440f0a | Tanguy Leroux | 12 April 2022, 10:11:12 UTC | Mute XPackRestIT deprecation/10_basic/Test Deprecations (#85807) (#85809) Relates #85806 | 12 April 2022, 10:11:12 UTC |
| 1ac3a71 | Przemysław Witek | 12 April 2022, 09:48:41 UTC | [Transform] Include missing privileges in the error message (#85547) (#85805) | 12 April 2022, 09:48:41 UTC |
| 51bb0b3 | Tanguy Leroux | 12 April 2022, 09:35:18 UTC | AwaitsFix for #85760 (#85810) Co-authored-by: David Turner <david.turner@elastic.co> | 12 April 2022, 09:35:18 UTC |
| eb00a14 | Salvatore Campagna | 12 April 2022, 08:49:02 UTC | Update v8.2.0 BC2 docs (#85771) | 12 April 2022, 08:49:02 UTC |
| 01635f3 | Yang Wang | 12 April 2022, 08:31:05 UTC | [Test] Use thread-safe hashSet for result collection (#85653) (#85804) Use regular HashSet in a deliberate concurrency code block sometimes results into unexpected behaviour: The hashSet has only a single item but size() reports 2. This PR replaces the regular hashSet with a thread-safe version to fix spurious test failures. Resolves: #85411 | 12 April 2022, 08:31:05 UTC |
| 0b0003f | Mark Vieira | 11 April 2022, 19:25:33 UTC | Update URL used for build scan CI metrics links | 11 April 2022, 19:26:54 UTC |
| 741982c | Tanguy Leroux | 11 April 2022, 11:58:29 UTC | Remove special rejection handling in Cold/Frozen cache services (#85775) (#85781) Now #85217 is merged the special handling introduced by #85239 to fail Gaps when a task is rejected can be removed. The `onRejection` method of the AbstractRunnable task will be executed if the task is rejected or if an unexpected exception is thrown before the task is added to the thread pool executor queue. | 11 April 2022, 11:58:29 UTC |
| 50f873a | Carlos Delgado | 11 April 2022, 08:56:17 UTC | Add indices permissions to Enterprise Search service account (#85726) (#85770) | 11 April 2022, 08:56:17 UTC |
| a65ee36 | David Turner | 07 April 2022, 16:40:20 UTC | Relax docs about using new repos across major versions (#85740) (#85754) Today the docs say that new major versions need new snapshot repositories or else corruption may occur. This isn't true, we support using the same repository across upgrades. | 07 April 2022, 16:40:20 UTC |
| 113bbb6 | István Zoltán Szabó | 07 April 2022, 08:28:27 UTC | [DOCS] Adds size param to evaluate DFA API docs (#85735) (#85736) | 07 April 2022, 08:28:27 UTC |
| 70ce232 | Dimitris Athanasiou | 06 April 2022, 13:57:08 UTC | [ML] Allow retrieving `boolean` fields from _source in DFA jobs (#85672) (#85724) This commit fixes a bug where `boolean` mapped fields may only be retrieved from doc values in Data Frame Analytics jobs. While this problem may occur for `boolean` fields with disabled doc values, a more common way of being affected by this bug is running an analysis against data with a very high number of fields. If there are more fields than the `index.max_docvalue_fields_search` setting, we fall back to retrieving the fields from _source. Then `boolean` mapped fields would fail with the error message "cannot apply boolean mapping to field [<field>]". | 06 April 2022, 13:57:08 UTC |
| 1ed69c1 | Martijn van Groningen | 06 April 2022, 11:11:23 UTC | Mute SnapshotLifecycleServiceTests#testPolicyCRUD() test. (#85715) (#85718) Relates to #44997 | 06 April 2022, 11:11:23 UTC |
| a70d75f | Salvatore Campagna | 06 April 2022, 08:17:40 UTC | Update docs for v8.2.0 release (#85664) | 06 April 2022, 08:17:40 UTC |
| 6254a71 | Mayya Sharipova | 05 April 2022, 18:10:35 UTC | Carry over release highlights from Lucene 9.1 (#85571) | 05 April 2022, 18:10:35 UTC |
| 6b34392 | Gordon Brown | 05 April 2022, 00:30:25 UTC | Enable overrides of license expiration dates (#85351) (#85685) In rare cases, we are required to adjust the expiration date of a license that already exists for legal or accounting reasons, rather than simply generating a new one. This commit introduces a basic mechanism for overriding licenses by hash of license ID, by way of a list currently hardcoded in the repo. This approach is fairly naive, and should be replaced as noted in the code if the frequency we are required to use this mechanism increases. This PR adds Java-based REST tests for the Put License API to help validate this behavior. | 05 April 2022, 00:30:25 UTC |
| f2ce727 | Joe Gallo | 04 April 2022, 21:26:52 UTC | Wait for tasks to finish after the forcemerge yml test (#85683) | 04 April 2022, 21:29:12 UTC |
| 8d17def | Joe Gallo | 04 April 2022, 17:43:50 UTC | Remove ILM docs redirect (#85628) | 04 April 2022, 17:44:43 UTC |
| 1fc8981 | Lee Hinman | 04 April 2022, 15:07:45 UTC | Bump commons-compress to 1.21 (#85581) (#85678) This updates the dependency version for the Apache commons compress project to 1.21 # Conflicts: # build-tools/build.gradle | 04 April 2022, 15:07:45 UTC |
| 1f0327c | Benjamin Trent | 04 April 2022, 12:45:20 UTC | [ML] ensure that inference index primary shards are available before attempting to load model (#85569) (#85674) when we attempt to load trained models nodes, we should wait for the inference indices primary shards to be available before attempting to load the models. Otherwise attempting to load the models is a waste of time and could cause errors. closes: #85438 | 04 April 2022, 12:45:20 UTC |
| e09cfe9 | Benjamin Trent | 04 April 2022, 12:37:52 UTC | [ML] ensure roberta tokenization options are updateable (#85578) (#85673) This ensures roberta tokenization options are updateable. The previous reberta code missed a case. Additionally, this commit improves testing robustness to avoid such a mistake again. | 04 April 2022, 12:37:52 UTC |
| 7eda06d | Francisco Fernández Castaño | 04 April 2022, 11:36:08 UTC | Ignore frozen shared cache file during data folder upgrades (#85666) Closes #85603 Backport of #85638 | 04 April 2022, 11:36:08 UTC |
| 0ba1042 | Nhat Nguyen | 01 April 2022, 22:04:48 UTC | Fix skip caching factor with indices.queries.cache.all_segments (#85510) (#85648) The indices.queries.cache.all_segments setting should not be used in production environments. And we should also use the default skipCachingFactor (i.e, 10) for when that setting is enabled. Co-authored-by: Elastic Machine <elasticmachine@users.noreply.github.com> | 01 April 2022, 22:04:48 UTC |
| 1b74395 | Nik Everett | 01 April 2022, 21:53:03 UTC | Fix filters agg test (#85650) (#85652) It was failing spuriously because it asserted that it could never use metadata to power the agg - but on some seed it'd get lucky and have an empty segment. Empty segments could use metadata. Which made the test fail. This makes sure there are not empty segments so we get the assertion we were looking for. Closes #85198 | 01 April 2022, 21:53:03 UTC |
| b6e695f | David Roberts | 01 April 2022, 19:13:34 UTC | [ML] Simplify shortcircuiting scores renormalizer (#85625) (#85643) In #85555 a Phaser was used to track ongoing renormalizations and wait for them to complete. However, this is a very complex way to do things. It turns out there's an easier way, as submitting a task to a threadpool returns a Future that can be used to track when the submitted task completes. There was also a bug in the ShortCircuitingRenormalizer class where the possibility of RejectedExecutionException was not being considered. | 01 April 2022, 19:13:34 UTC |
| d92ec7a | Mayya Sharipova | 01 April 2022, 18:49:30 UTC | Describe release highlight for ANN filtering (#85622) | 01 April 2022, 18:50:55 UTC |
| 187dfcc | William Brafford | 01 April 2022, 18:42:29 UTC | Log at WARN level for Watcher cluster state validation errors (#85632) (#85642) * Log at WARN level for Watcher cluster state validation errors * Update docs/changelog/85632.yaml | 01 April 2022, 18:42:29 UTC |
| 9aaa015 | Nikola Grcevski | 01 April 2022, 17:05:12 UTC | [TEST] Fix test failures on Windows (#85498) (#85633) * Fix assert matching for Windows. * Disable Windows incompatible tests | 01 April 2022, 17:05:12 UTC |
| 74a4ffa | Keith Massey | 01 April 2022, 16:45:52 UTC | Avoiding watcher validation errors when a data stream points to more than one index (#85507) (#85631) This commit fixes a validation bug that prevented watcher from starting if the .watcher-history-16 data stream is pointing to more than one index. Before 8.0, .watcher-history-16 data had been an alias and had never pointed to more than one index, so this had not been a problem. | 01 April 2022, 16:45:52 UTC |
| 1023ca5 | David Roberts | 01 April 2022, 15:27:41 UTC | [ML] Avoid multiple queued quantiles documents in renormalizer (#85555) (#85597) It's possible that an anomaly detection job can output a flurry of quantiles documents for use in renormalization in the time that a single renormalization takes. When this happens we only use the most recent quantiles document for the next renormalization, but the previous code queued up all the quantiles documents until the next renormalization started. The new approach that this PR implements is to only keep the most recent new quantiles document while the current renormalization is in progress. This avoids wasting memory on a queue that will be largely discarded (apart from one element) later on. Fixes #85539 | 01 April 2022, 15:27:41 UTC |
| 0a96907 | Przemyslaw Gomulka | 01 April 2022, 13:36:03 UTC | mute PackageUpgradeTests.test21CheckUpgradedVersion (#85616) (#85618) relates #85386 | 01 April 2022, 13:36:03 UTC |
| 80cd2ed | Ignacio Vera | 01 April 2022, 12:56:16 UTC | Fix fields wildcard support to vector tile search API (#85595) (#85612) This commit changes the strategy on how fields are fetched by iterating over the returned map. In addition, the geo field is added the last one on the fields list so it does not get overridden. | 01 April 2022, 12:56:16 UTC |
| 2b4845f | Rene Groeschke | 01 April 2022, 12:05:40 UTC | Update gradle wrapper to 7.4.2 (#85608) (#85611) | 01 April 2022, 12:05:40 UTC |
| 266d095 | Yannick Welsch | 01 April 2022, 07:21:51 UTC | [8.2] Fix wildcard highlighting on match_only_text (#85500) (#85594) Fixes a bug where match_only_text fields were ignored during highlighting when a field name with wildcard was specified. Closes #85493 | 01 April 2022, 07:21:51 UTC |
| ccfc925 | Tim Vernum | 01 April 2022, 02:23:25 UTC | Ignore app priv failures when resolving superuser (#85519) (#85587) In #81400 we changed `superuser` to no longer have _every_ privilege. Consequently, we also removed the special case code that existed that would ignore all other roles for any user that had superuser role. However, we added some special handling so that failing to resolve those other roles would not block superuser access - when a user has superuser role, any failures in role resolution will be effectively ignored, and the user will be given the superuser role only. However, this failure handling did not account for the loading of application privileges. If application privileges needed to be loaded, but failed, this could prevent resolution of the superuser role. This change extends the failure handling to encompass the full resolution of roles, and fallback to superuser only, whenever other roles or application privileges are unavailable Relates: #85312 | 01 April 2022, 02:23:25 UTC |
| 78d76e0 | Yang Wang | 01 April 2022, 00:33:14 UTC | User Profile - rename search to suggest profile (#85196) (#85584) This PR renames "search profile" to "suggest profile" in all relevant places: * Class names * Variable names * Docs * URL and REST specs | 01 April 2022, 00:33:14 UTC |
| 49dba59 | Yang Wang | 31 March 2022, 23:43:59 UTC | [Test] Add more logging for testConcurrentCreationOfNewProfiles (#85521) (#85583) This PR adds more logging to the test so that test failure will likely include more information to help diagnosis. Relates: #85411 | 31 March 2022, 23:43:59 UTC |
| 67aedb0 | Ryan Ernst | 31 March 2022, 22:00:52 UTC | Move transform schema tests to transform module (#85554) (#85564) The x-pack transform module has several tests related to schmea validation. This commit moves those tests from x-pack core into the module itself. The :test:x-content dependency is then isolated to transform, so that it does not affect other x-pack modules that do not need it, and that could conflict with the runtime jackson dependency it brings in. | 31 March 2022, 22:00:52 UTC |
| b4097b6 | Artem Prigoda | 31 March 2022, 19:09:01 UTC | Bump versions for 8.1.2 release | 31 March 2022, 19:10:26 UTC |
| a9145bb | Artem Prigoda | 31 March 2022, 18:53:29 UTC | Bump versions for 7.17.2 release | 31 March 2022, 19:01:38 UTC |
| 184ace7 | Joe Gallo | 31 March 2022, 17:43:57 UTC | Remove "Push back excessive requests for stats (#83832)" from 8.2 branch (#85504) | 31 March 2022, 17:43:57 UTC |
| b6d35cf | Salvatore Campagna | 31 March 2022, 15:55:52 UTC | fix: make EmptySortValue instances sort after everything else (#85437) (#85562) When comparing instances of SortValue we make sure that instances of subclass ValuelessSortValue sort after everything else. This is also to be consistent with the way Double.NaN values are sorted with respect to Double values. The name `valueless` is chosen so that instances of ValuelessSortValue sorts after `bytes`, `double` and `long`, the names of the other three existing subclasses. Additionally we make sure to create an instance of ValuelessSortValue if creating an instance from a Double.NaN value. This PR is intended to fix a potential issue introduced by #85058. | 31 March 2022, 15:55:52 UTC |
| e1c53ea | Bogdan Pintea | 31 March 2022, 15:02:08 UTC | SQL: Update DbVisualizer connection steps (#85167) (#85561) Update DbVisualizer connection steps to reflect version 13.0's integrated Elasticsearch support. | 31 March 2022, 15:02:08 UTC |
| 66fc693 | Artem Prigoda | 31 March 2022, 14:44:07 UTC | DOCS: [8.2] Add 8.1.2 release notes (#85448) (#85558) | 31 March 2022, 14:44:07 UTC |
| 0e4d103 | Lukas Wegmann | 31 March 2022, 14:29:25 UTC | Fix SQLCompatIT.testCursorFromOldNodeFailsOnNewNode (#85531) (#85549) resolves #85520 The failure was caused by the assumption that groupBy queries always return a scroll cursor that has been fixed in https://github.com/elastic/elasticsearch/pull/84356. Because we will run into the same problem again with 8.2.1 this fix also needs to be backported to 8.2. | 31 March 2022, 14:29:25 UTC |
| 25c4650 | Benjamin Trent | 31 March 2022, 13:52:37 UTC | [ML] fix transform privilege requirements documentation (#85544) (#85552) This updates the privileges required when a retention_policy is configured. See related PR: #85413 | 31 March 2022, 13:52:37 UTC |
| e34601d | Yannick Welsch | 31 March 2022, 13:41:59 UTC | Remove archival functionality from 8.2 branch (#85524) Archival won't ship with 8.2, hence removing the files there. I found it easier to remove the code than reintroducing a way to only enable it in snapshot builds. Relates #81210 | 31 March 2022, 13:41:59 UTC |
| f25346d | David Turner | 31 March 2022, 08:12:16 UTC | Add technical docs on diagnosing instability etc (#85074) (#85528) Copies some internal troubleshooting docs to the reference manual for wider use. Co-authored-by: James Rodewig <james.rodewig@gmail.com> Co-authored-by: James Rodewig <james.rodewig@gmail.com> | 31 March 2022, 08:12:16 UTC |
| d2d8757 | Mark Vieira | 30 March 2022, 14:43:44 UTC | Don't run PR checks on feature/desired-balance-allocator branch | 30 March 2022, 14:43:56 UTC |
| be878c3 | Martijn van Groningen | 30 March 2022, 14:21:28 UTC | Remove HLRC IngestClient class. (#85483) including GeoIpStatsResponse and IngestRequestConverters classes. This one was easy as there were no usages of this HLRC sub client. Relates to #83423 | 30 March 2022, 14:21:28 UTC |
| 5da038b | Lee Hinman | 30 March 2022, 14:15:22 UTC | Remove HLRC's SearchableSnapshotsClient (#85472) This commit removes this sub-client as part of the work to remove the HLRC entirely. Relates to #83423 | 30 March 2022, 14:15:22 UTC |
| 42b0b39 | Joe Gallo | 30 March 2022, 13:13:16 UTC | Fix EnrichSecurityIT resource cleanup (#85469) | 30 March 2022, 13:13:16 UTC |
| b1b54d5 | Salvatore Campagna | 30 March 2022, 12:51:03 UTC | test: date_histogram with time zone on date_nanos field (#85149) | 30 March 2022, 12:51:03 UTC |
| 0b84eb1 | Craig Taverner | 30 March 2022, 12:29:01 UTC | Added buffer to vector tile REST API docs (#85460) | 30 March 2022, 12:29:01 UTC |
| d95e76b | Benjamin Trent | 30 March 2022, 12:24:27 UTC | [ML] disallow new trained model deployments when nodes are different versions (#85465) Since trained model deployments are experimental. Backwards compatibility is not always guaranteed between versions. To protect users, this commit disallows new deployments to be created in a mixed versioned cluster environment. | 30 March 2022, 12:24:27 UTC |
| d36d7c2 | Milton Hultgren | 30 March 2022, 11:24:55 UTC | Disable dynamic mapping in Stack Monitoring index templates (#85447) This aligns the new Metricbeat templates with the settings found in the older internal collection templates. By disabling dynamic mapping we avoid the possible field explosion that can happen for example when there are lots of nodes in the Elasticsearch cluster stats response which could easily blow past the 2 000 field limit. This commit also adds the missing ILM policy to the Enterprise search mappings. | 30 March 2022, 11:24:55 UTC |
| 2d85fa7 | Rene Groeschke | 30 March 2022, 09:08:34 UTC | Update build convention dependencies (#85444) - Update spotless to potentially fix invalid git worktree handling - Update shadow plugin to be in sync with version used in build-tools | 30 March 2022, 09:08:34 UTC |
| 087bd1d | Luigi Dell'Aquila | 30 March 2022, 06:50:43 UTC | SQL: Improve ROUND and TRUNCATE functions - preserve input type (eg. ROUND(Long) returns a Long, ROUND(Double) returns Double) - avoid loss of precision and number overflows in the calculation algorithm - improve performance on common cases (eg. ROUND(N, 0), ROUND(N, 2) ...) | 30 March 2022, 06:50:43 UTC |
| 92c4538 | Ievgen Degtiarenko | 30 March 2022, 06:18:26 UTC | Ensure the cleanup in testAutoFollowPatterns is always happening (#85401) In case of failure testAutoFollowPatterns will leave auto-following pattern running that would case assertion failure in other test cases | 30 March 2022, 06:18:26 UTC |
| 105cdeb | Ryan Ernst | 30 March 2022, 01:44:32 UTC | Use Set instead of LongSet in long script field (#85417) The long script field uses hppc's LongSet to store the unique values to be queried. However, this set should be relatively small. This commit changes the internals of the runtime long script field to use Set<Long>. relates #84735 | 30 March 2022, 01:44:32 UTC |
| 53fb1d8 | Ryan Ernst | 30 March 2022, 01:42:38 UTC | Remove hppc from search tests (#85468) This commit removes uses of hppc collections in search tests that are purely for local record keeping. relates #84735 | 30 March 2022, 01:42:38 UTC |
| 898d849 | Armin Braun | 29 March 2022, 22:21:56 UTC | Make classes+methods that can be static static in many spots (#85370) Just some quick static analysis+fixing here. Not much in terms of code changes besides adding the `static` keywords with the exception of some simplifications to some of the search objects that don't need the search controller instance passed down in many spots. This was done mostly automatically by the IDE but some quick manual inspection shows quite a few spots where this should make things behave better via things like making lambdas non-capturing. | 29 March 2022, 22:21:56 UTC |
| ee864bd | Mark Vieira | 29 March 2022, 21:56:14 UTC | Mute NodeShutdownIT.testShutdownReadinessService | 29 March 2022, 21:56:14 UTC |
| 4d189f6 | Luca Cavanna | 29 March 2022, 18:01:29 UTC | Remove MappingLookup#isShadowed (#84991) With #75454 we have fixed a buggy behaviour that was causing us to treat multi-fields as if they were runtime fields when retrieving mappers for the fields encountered while parsing incoming documents. As a follow-up #75595 introduced MappingLookup#isShadowed as a primitive to check if a field is a runtime field which covers also for subfields of the runtime composite field which the previous check did not. With recent changes around expanding dots in field names, the situation where we previously ended up confusing a multi-field for a runtime field no longer manifests.This allows us to potentially remove isShadowed and go back to the original logic + assertion: if we could not resolve the field to a mapper by name, and we can resolve it to a field type by name, such field must be a runtime field. This is maybe not as elegant as isShadowed but I found its naming a bit confusing and I was hoping we can avoid maintaining a specialized method around this. | 29 March 2022, 18:01:29 UTC |
| 3be4d6b | Salvatore Campagna | 29 March 2022, 17:56:16 UTC | test: timezone with data_histogram and composite date_histogram (#85176) | 29 March 2022, 17:56:16 UTC |
| dacb9a1 | Joe Gallo | 29 March 2022, 17:44:47 UTC | Allowing _cat/allocation to return more than 0 shards in test (#85459) | 29 March 2022, 17:44:47 UTC |
| 50d1b4f | Artem Prigoda | 29 March 2022, 17:41:13 UTC | Do not fail on duplicated content field filters (#85382) `Set.of()` throws a runtime exception if it encounters a duplicated element, `Set.copyOf`, on the contrary, doesn't do that. We shouldn't fail if the user configure multiple include or exclude filters for _source fields. See https://www.elastic.co/guide/en/elasticsearch/reference/current/mapping-source-field.html#include-exclude | 29 March 2022, 17:41:13 UTC |
| 9071f59 | Nik Everett | 29 March 2022, 16:32:09 UTC | Remove duplicate runtime fields tests (#85453) These are duplicated with tests in `modules/runtime-fields-common`. Closes #85452 | 29 March 2022, 16:32:09 UTC |
| 9a60aa8 | Craig Taverner | 29 March 2022, 15:42:45 UTC | Support vector tile buffer in REST API (#85450) * Support vector tile buffer in REST API In a previous PR we supported the configurable buffer for field formatting in the mvt() function, but not in the REST API. This commit adds the missing piece. * Added basic yamlRestTest for buffer param | 29 March 2022, 15:42:45 UTC |
| c9ed0c0 | Armin Braun | 29 March 2022, 15:21:39 UTC | Extract Cold Paths in DocumentParser (#85268) Looking at the profiling of the indexing steps in the many-shards benchmarks it seems like we don't inline much in these nested parser calls and spent quite a bit of time on method calls. Lets give the compiler an easier time by extracting the cold exception paths out of this logic. | 29 March 2022, 15:21:39 UTC |
| 42d0071 | Lee Hinman | 29 March 2022, 15:02:26 UTC | Correct prefix in docs for ILM fully-mounted searchable snapshots (#85457) This incorrectly had "recovered-" when "restored-" is what is actually used. Resolves #85432 | 29 March 2022, 15:02:26 UTC |
