[
https://issues.apache.org/jira/browse/IGNITE-9787?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16839305#comment-16839305
]
Alexey Goncharuk commented on IGNITE-9787:
------------------------------------------
Looks like the cause in
{{org.apache.ignite.internal.processors.cache.query.GridCacheQueryAdapter#executeScanQuery}}.
If {{part == null}}, then no partition state check will be executed.
The most logical way to fix this is to make full scan query to scan partitions
one-by-one.
> Local Continuous Query skips rows during initial rebalance of REPLICATED cache
> ------------------------------------------------------------------------------
>
> Key: IGNITE-9787
> URL: https://issues.apache.org/jira/browse/IGNITE-9787
> Project: Ignite
> Issue Type: Bug
> Affects Versions: 2.4
> Reporter: Ilya Kasnacheev
> Priority: Major
> Attachments: IgniteRepro2.zip
>
>
> This is observer under .Net but I assume will affect Java too:
> If you have a partitioned cache with non-local continuous query, continuous
> query will iterate every entry in REPLICATED cache.
> If you have a replicated cache with non-local continuous query, it works.
> If you have a replicated cache with local continuous query and synchronous
> rebalancing, it also works.
> But if you have replicated cache with local continuous query and
> non-synchronous rebalancing , you will see the gap in data - newly written
> entries while rebalance is underway may not be passed to callback or scan
> query.
> Please see the ML thread and reproducer.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
