Add sequence numbers based optimistic concurrency control support to Engine by bleskes · Pull Request #36467 · elastic/elasticsearch

bleskes · 2018-12-11T09:04:21Z

This PR add support to engine operations for resolving and verifying the sequence number and primary term of the last modification to a document before performing an operation. This is infrastructure to move our optimistic concurrency control API to use sequence numbers instead of internal versioning.

Relates #36148
Relates #10708

PS there are still discussions about the parameter naming. I'll rename these once we have settled on something.

elasticmachine · 2018-12-11T09:04:23Z

Pinging @elastic/es-distributed

s1monw

LGTM - I am not happy with the naming. I think cas should not be in there. Can we just name them expectedSeqId expectedPrimaryTerm for now? That's what for instance Integer.compareAndSet uses as it's parameter names.

s1monw · 2018-12-11T20:51:26Z

server/src/main/java/org/elasticsearch/common/lucene/uid/PerThreadIDVersionAndSeqNoLookup.java

+            final long seqNo;
+            final long term;
+            if (loadSeqNo) {
+                NumericDocValues seqNos = context.reader().getNumericDocValues(SeqNoFieldMapper.NAME);


we should clean this up in master once backported. can you put a comment in there?

dnhatn

LGTM.

server/src/test/java/org/elasticsearch/index/engine/InternalEngineTests.java

bleskes · 2018-12-12T12:04:37Z

Thanks @s1monw and @dnhatn. I incorporated your suggestion.

bleskes · 2018-12-12T16:55:37Z

run gradle build tests 1

bleskes · 2018-12-12T16:59:58Z

The build failure was regarding refresh listeners. I don't think it's a related but probably signals a concurrency bug. I tried to reproduce this without success (I run ~300 time). Capturing the info here in case it's needed.

15:42:40 FAILURE 18.0s J0 | RefreshListenersTests.testConcurrentRefresh <<< FAILURES!
15:42:40    > Throwable #1: java.lang.AssertionError
15:42:40    > 	at __randomizedtesting.SeedInfo.seed([2848E94A583B5A6E:EF0293D9D997F804]:0)
15:42:40    > 	at org.elasticsearch.index.shard.RefreshListenersTests.lambda$testConcurrentRefresh$8(RefreshListenersTests.java:282)
15:42:40    > 	at org.elasticsearch.test.ESTestCase.assertBusy(ESTestCase.java:847)
15:42:40    > 	at org.elasticsearch.test.ESTestCase.assertBusy(ESTestCase.java:821)
15:42:40    > 	at org.elasticsearch.index.shard.RefreshListenersTests.testConcurrentRefresh(RefreshListenersTests.java:282)

./gradlew :server:test -Dtests.seed=2848E94A583B5A6E -Dtests.class=org.elasticsearch.index.shard.RefreshListenersTests -Dtests.method="testConcurrentRefresh" -Dtests.security.manager=true -Dtests.locale=pt-PT -Dtests.timezone=UCT -Dcompiler.java=11 -Druntime.java=8

ywelsch · 2018-12-12T17:06:44Z

The build failure was regarding refresh listeners. I don't think it's a related

Similar assertion recently failed on master (#24418 (comment)), so looks to be unrelated to your changes here

bleskes · 2018-12-12T17:22:35Z

run gradle build tests 1

…Engine (elastic#36467) This commit add support to engine operations for resolving and verifying the sequence number and primary term of the last modification to a document before performing an operation. This is infrastructure to move our (optimistic concurrency control)[http://en.wikipedia.org/wiki/Optimistic_concurrency_control] API to use sequence numbers instead of internal versioning. Relates elastic#36148 Relates elastic#10708

…Engine (#36467) This commit add support to engine operations for resolving and verifying the sequence number and primary term of the last modification to a document before performing an operation. This is infrastructure to move our (optimistic concurrency control)[http://en.wikipedia.org/wiki/Optimistic_concurrency_control] API to use sequence numbers instead of internal versioning. Relates #36148 Relates #10708

Changes were made in #34357 and #36467

The assertion `assertOpsOnPrimary` does not store seq_no and primary term of successful deletes to the `lastOpSeqNo` and `lastOpTerm`. This leads to failures of the subsequence CAS deletes or indexes with seq_no and term. Moreover, this assertion trips a translog assertion because it bumps the primary term of some operations but not the primary term of the engine. Relates elastic#36467 Closes elastic#37684

The assertion `assertOpsOnPrimary` does not store seq_no and primary term of successful deletes to the `lastOpSeqNo` and `lastOpTerm`. This leads to failures of the subsequence CAS deletes or indexes with seq_no and term. Moreover, this assertion trips a translog assertion because it bumps the primary term of some operations but not the primary term of the engine. Relates #36467 Closes #37684

Changes were made in #34357 and #36467

The assertion `assertOpsOnPrimary` does not store seq_no and primary term of successful deletes to the `lastOpSeqNo` and `lastOpTerm`. This leads to failures of the subsequence CAS deletes or indexes with seq_no and term. Moreover, this assertion trips a translog assertion because it bumps the primary term of some operations but not the primary term of the engine. Relates elastic#36467 Closes elastic#37684

The assertion `assertOpsOnPrimary` does not store seq_no and primary term of successful deletes to the `lastOpSeqNo` and `lastOpTerm`. This leads to failures of the subsequence CAS deletes or indexes with seq_no and term. Moreover, this assertion trips a translog assertion because it bumps the primary term of some operations but not the primary term of the engine. Relates #36467 Closes #37684

Changes were made in elastic#34357 and elastic#36467

bleskes added 4 commits December 7, 2018 16:26

add seq no powered cas to engine and friends.

4343a83

merge from master

59ef127

Merge remote-tracking branch 'upstream/master' into cas_seq_no_engine

fe89a07

assert messages

1782455

bleskes added >enhancement v7.0.0 :Distributed/Engine Anything around managing Lucene and the Translog in an open shard. v6.6.0 labels Dec 11, 2018

bleskes requested review from dnhatn, s1monw and ywelsch December 11, 2018 09:04

s1monw approved these changes Dec 11, 2018

View reviewed changes

dnhatn approved these changes Dec 11, 2018

View reviewed changes

server/src/test/java/org/elasticsearch/index/engine/InternalEngineTests.java Outdated Show resolved Hide resolved

bleskes added 2 commits December 12, 2018 12:11

Merge remote-tracking branch 'upstream/master' into cas_seq_no_engine

fe9a6f8

feedback

365d56a

moved to if*Match terminology

594bf52

bleskes force-pushed the cas_seq_no_engine branch from 014528f to 594bf52 Compare December 12, 2018 15:27

bleskes merged commit f6b5d7e into elastic:master Dec 13, 2018

bleskes deleted the cas_seq_no_engine branch December 13, 2018 07:08

tlrx added a commit that referenced this pull request Jan 9, 2019

[RCI] Adapt NoOpEngine to latest FrozenEngine changes

b013d49

Changes were made in #34357 and #36467

tlrx added a commit that referenced this pull request Jan 10, 2019

[RCI] Adapt NoOpEngine to latest FrozenEngine changes

a7d0731

Changes were made in #34357 and #36467

tlrx added a commit that referenced this pull request Jan 11, 2019

[RCI] Adapt NoOpEngine to latest FrozenEngine changes

269cb0b

Changes were made in #34357 and #36467

tlrx added a commit that referenced this pull request Jan 15, 2019

[RCI] Adapt NoOpEngine to latest FrozenEngine changes

fbbeff4

Changes were made in #34357 and #36467

tlrx added a commit that referenced this pull request Jan 22, 2019

[RCI] Adapt NoOpEngine to latest FrozenEngine changes

b00b323

Changes were made in #34357 and #36467

dnhatn mentioned this pull request Jan 23, 2019

Fix InternalEngineTests#assertOpsOnPrimary #37746

Merged

tlrx added a commit that referenced this pull request Jan 29, 2019

[RCI] Adapt NoOpEngine to latest FrozenEngine changes

1e55f44

Changes were made in #34357 and #36467

tlrx added a commit that referenced this pull request Jan 29, 2019

[RCI] Adapt NoOpEngine to latest FrozenEngine changes

c3a8ead

Changes were made in #34357 and #36467

tlrx added a commit that referenced this pull request Jan 30, 2019

[RCI] Adapt NoOpEngine to latest FrozenEngine changes

54d110b

Changes were made in #34357 and #36467

colings86 added v7.0.0-beta1 and removed v7.0.0 labels Feb 7, 2019

talevy mentioned this pull request Feb 15, 2019

Fix InternalEngineTests#assertOpsOnPrimary (#37746) #38990

Merged

dnhatn mentioned this pull request Feb 27, 2019

Use sequences numbers for Compare and Set operations #26493

Closed

tlrx added a commit to tlrx/elasticsearch that referenced this pull request Mar 1, 2019

[RCI] Adapt NoOpEngine to latest FrozenEngine changes

9eab2c7

Changes were made in elastic#34357 and elastic#36467

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add sequence numbers based optimistic concurrency control support to Engine#36467

Add sequence numbers based optimistic concurrency control support to Engine#36467
bleskes merged 7 commits intoelastic:masterfrom
bleskes:cas_seq_no_engine

bleskes commented Dec 11, 2018 •

edited

Loading

Uh oh!

elasticmachine commented Dec 11, 2018

Uh oh!

s1monw left a comment

Uh oh!

s1monw Dec 11, 2018

Uh oh!

dnhatn left a comment

Uh oh!

Uh oh!

bleskes commented Dec 12, 2018

Uh oh!

bleskes commented Dec 12, 2018

Uh oh!

bleskes commented Dec 12, 2018

Uh oh!

ywelsch commented Dec 12, 2018

Uh oh!

bleskes commented Dec 12, 2018

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

Conversation

bleskes commented Dec 11, 2018 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

elasticmachine commented Dec 11, 2018

Uh oh!

s1monw left a comment

Choose a reason for hiding this comment

Uh oh!

s1monw Dec 11, 2018

Choose a reason for hiding this comment

Uh oh!

dnhatn left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

bleskes commented Dec 12, 2018

Uh oh!

bleskes commented Dec 12, 2018

Uh oh!

bleskes commented Dec 12, 2018

Uh oh!

ywelsch commented Dec 12, 2018

Uh oh!

bleskes commented Dec 12, 2018

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

6 participants

bleskes commented Dec 11, 2018 •

edited

Loading