[
https://issues.apache.org/jira/browse/SOLR-14923?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17254315#comment-17254315
]
David Smiley commented on SOLR-14923:
-------------------------------------
Today I made further modifications to NestedShardedAtomicUpdateTest so that it
didn't always call /get to make some assertions on the results -- I guarded
this with a random boolean. I'm aware that /get can _sometimes_ trigger a new
realtime searcher, and so I wanted to test that the side effect of this wasn't
masking bugs. It did find a bug :-(. And I tracked it down; it's limited to
in-place-updates of a child doc because these updates are added to the update
log ID'ed by the document itself, not the root ID it is a part of.
Consequently, a /get later of the root ID asking for child docs (fl=*,[child])
can slip past the updateLog (because the ulog has no updates of this root ID)
and the _existing_ realtimeSearcher will be used, returning a stale value of
the updated child document. This is despite {{mustUseRealtimeSearcher==true}}
because that only matters if a doc is found in the updateLog. RTG goes through
some great lengths (with added complexity thereof) to avoid calling
{{ulog.openRealtimeSearcher}}.
I've tried some ideas and I'll think more on a fix for this one.
> Indexing performance is unacceptable when child documents are involved
> ----------------------------------------------------------------------
>
> Key: SOLR-14923
> URL: https://issues.apache.org/jira/browse/SOLR-14923
> Project: Solr
> Issue Type: Bug
> Security Level: Public(Default Security Level. Issues are Public)
> Components: update, UpdateRequestProcessors
> Affects Versions: 8.3, 8.4, 8.5, 8.6, 8.7, master (9.0)
> Reporter: Thomas Wöckinger
> Priority: Critical
> Labels: performance, pull-request-available
> Time Spent: 1h
> Remaining Estimate: 0h
>
> Parallel indexing does not make sense at moment when child documents are used.
> The org.apache.solr.update.processor.DistributedUpdateProcessor checks at the
> end of the method doVersionAdd if Ulog caches should be refreshed.
> This check will return true if any child document is included in the
> AddUpdateCommand.
> If so ulog.openRealtimeSearcher(); is called, this call is very expensive,
> and executed in a synchronized block of the UpdateLog instance, therefore all
> other operations on the UpdateLog are blocked too.
> Because every important UpdateLog method (add, delete, ...) is done using a
> synchronized block almost each operation is blocked.
> This reduces multi threaded index update to a single thread behavior.
> The described behavior is not depending on any option of the UpdateRequest,
> so it does not make any difference if 'waitFlush', 'waitSearcher' or
> 'softCommit' is true or false.
> The described behavior makes the usage of ChildDocuments useless, because the
> performance is unacceptable.
>
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org
For additional commands, e-mail: issues-h...@lucene.apache.org
