[
https://issues.apache.org/jira/browse/IGNITE-10218?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16683539#comment-16683539
]
Sergey Antonov commented on IGNITE-10218:
-----------------------------------------
[~sboikov], could we know your opinion?
> Detecting lost partitions PME phase triggered twice on coordinator
> ------------------------------------------------------------------
>
> Key: IGNITE-10218
> URL: https://issues.apache.org/jira/browse/IGNITE-10218
> Project: Ignite
> Issue Type: Bug
> Reporter: Sergey Antonov
> Priority: Major
> Fix For: 2.8
>
>
> scenarion: 1 server left
> coordinator node seems to detect partition losses twice per exchange
> {noformat}
> [16:54:22,027][INFO][exchange-worker-#66][time] Finished exchange init
> [topVer=AffinityTopologyVersion [topVer=13, minorTopVer=0], crd=true]
> [16:54:22,163][INFO][db-checkpoint-thread-#86][GridCacheDatabaseSharedManager]
> Skipping checkpoint (no pages were modified) [checkpointLockWait=0ms,
> checkpointLockHoldTime=525ms, reason='timeout']
> [16:54:22,338][INFO][sys-#136][GridDhtPartitionsExchangeFuture] Coordinator
> received single message [ver=AffinityTopologyVersion [topVer=13,
> minorTopVer=0], node=2b69b32f-1bea-4c83-a70d-d7ff8ad7e319, allReceived=false]
> [16:54:22,401][INFO][sys-#137][GridDhtPartitionsExchangeFuture] Coordinator
> received single message [ver=AffinityTopologyVersion [topVer=13,
> minorTopVer=0], node=933628df-5237-435c-81d3-7d4be20d8cea, allReceived=false]
> [16:54:22,405][INFO][sys-#73][GridDhtPartitionsExchangeFuture] Coordinator
> received single message [ver=AffinityTopologyVersion [topVer=13,
> minorTopVer=0], node=549935a6-48b0-47cd-a763-13cef4706960, allReceived=false]
> [16:54:22,413][INFO][sys-#121][GridDhtPartitionsExchangeFuture] Coordinator
> received single message [ver=AffinityTopologyVersion [topVer=13,
> minorTopVer=0], node=fbdde4e1-2422-49af-a0bb-d6797cc723fe, allReceived=false]
> [16:54:22,722][INFO][sys-#122][GridDhtPartitionsExchangeFuture] Coordinator
> received single message [ver=AffinityTopologyVersion [topVer=13,
> minorTopVer=0], node=d67a748d-9c63-4ede-8dae-064b63dd1586, allReceived=true]
> [16:54:23,493][INFO][db-checkpoint-thread-#86][GridCacheDatabaseSharedManager]
> Skipping checkpoint (no pages were modified) [checkpointLockWait=0ms,
> checkpointLockHoldTime=849ms, reason='timeout']
> [16:54:23,494][INFO][sys-#122][GridDhtPartitionsExchangeFuture] Coordinator
> received all messages, try merge [ver=AffinityTopologyVersion [topVer=13,
> minorTopVer=0]]
> [16:54:23,494][INFO][sys-#122][GridDhtPartitionsExchangeFuture] Exchanges
> merging performed in 0 ms.
> [16:54:23,494][INFO][sys-#122][GridDhtPartitionsExchangeFuture]
> finishExchangeOnCoordinator [topVer=AffinityTopologyVersion [topVer=13,
> minorTopVer=0], resVer=AffinityTopologyVersion [topVer=13, minorTopVer=0]]
> [16:54:24,223][INFO][sys-#122][CacheAffinitySharedManager] Affinity
> recalculation (on server left) performed in 729 ms.
> [16:54:24,371][INFO][db-checkpoint-thread-#86][GridCacheDatabaseSharedManager]
> Skipping checkpoint (no pages were modified) [checkpointLockWait=0ms,
> checkpointLockHoldTime=726ms, reason='timeout']
> [16:54:25,146][INFO][db-checkpoint-thread-#86][GridCacheDatabaseSharedManager]
> Skipping checkpoint (no pages were modified) [checkpointLockWait=1ms,
> checkpointLockHoldTime=493ms, reason='timeout']
> [16:54:26,443][INFO][db-checkpoint-thread-#86][GridCacheDatabaseSharedManager]
> Skipping checkpoint (no pages were modified) [checkpointLockWait=8ms,
> checkpointLockHoldTime=776ms, reason='timeout']
> [16:54:26,443][INFO][sys-#122][GridDhtPartitionsExchangeFuture] Affinity
> changes (coordinator) applied in 2949 ms.
> [16:54:26,758][INFO][sys-#122][GridDhtPartitionsExchangeFuture] Partitions
> validation performed in 307 ms.
> [16:54:27,398][INFO][db-checkpoint-thread-#86][GridCacheDatabaseSharedManager]
> Skipping checkpoint (no pages were modified) [checkpointLockWait=0ms,
> checkpointLockHoldTime=725ms, reason='timeout']
> [16:54:27,646][INFO][sys-#122][GridDhtPartitionsExchangeFuture] Detecting
> lost partitions performed in 887 ms.
> [16:54:28,908][INFO][sys-#122][GridDhtPartitionsExchangeFuture] Preparing
> Full Message performed in 1138 ms.
> [16:54:28,908][INFO][sys-#122][GridDhtPartitionsExchangeFuture] Sending Full
> Message performed in 0 ms.
> [16:54:28,908][INFO][sys-#122][GridDhtPartitionsExchangeFuture] Sending Full
> Message to all nodes performed in 0 ms.
> [16:54:28,908][INFO][sys-#122][GridDhtPartitionsExchangeFuture] Finish
> exchange future [startVer=AffinityTopologyVersion [topVer=13, minorTopVer=0],
> resVer=AffinityTopologyVersion [topVer=13, minorTopVer=0], err=null]
> [16:54:29,171][INFO][db-checkpoint-thread-#86][GridCacheDatabaseSharedManager]
> Skipping checkpoint (no pages were modified) [checkpointLockWait=0ms,
> checkpointLockHoldTime=486ms, reason='timeout']
> [16:54:29,316][INFO][sys-#122][GridDhtPartitionsExchangeFuture] Detecting
> lost partitions performed in 407 ms.
> {noformat}
> This method is invoked two times:
> # GridDhtPartitionsExchangeFuture#finishExchangeOnCoordinator()
> # GridDhtPartitionsExchangeFuture#onDone()
> Do we really need to perform Detecting lost partitions twice?
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
