[ 
https://issues.apache.org/jira/browse/IGNITE-23584?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17894808#comment-17894808
 ] 

Vladimir Steshin commented on IGNITE-23584:
-------------------------------------------

[~zstan], thanks for notifying. Actually, the previous code worked correctly. 
But it looked really confusing. We decided to improve the semantics.

> Snapshot. Fix logic of the quick snapshot handler.
> --------------------------------------------------
>
>                 Key: IGNITE-23584
>                 URL: https://issues.apache.org/jira/browse/IGNITE-23584
>             Project: Ignite
>          Issue Type: Improvement
>    Affects Versions: 2.14
>            Reporter: Vladimir Steshin
>            Assignee: Vladimir Steshin
>            Priority: Trivial
>             Fix For: 2.17
>
>          Time Spent: 20m
>  Remaining Estimate: 0h
>
> _SnapshotPartitionsQuickVerifyHandler_ looks confusing. It continues checking 
> partitions even if detects the `no-need-to-work` condition and then skips the 
> collected results. We should not do any checks is the streamer warning is 
> detected. 
> {code:java}
> for (SnapshotHandlerResult<Map<PartitionKeyV2, PartitionHashRecordV2>> result 
> : results) {
>             if (result.error() != null)
>                 throw new IgniteCheckedException(result.error());
>             if (result.data() == null) {
>                 noData = true;
>                 continue;
>             }
>             Map<PartitionKeyV2, PartitionHashRecordV2> partsData = 
> result.data();
>             partsData.forEach((part, val) -> {
>                 PartitionHashRecordV2 other = total.putIfAbsent(part, val);
>                 if ((other != null && !wrnGrps.contains(part.groupId()))
>                     && ((!val.hasExpiringEntries() && 
> !other.hasExpiringEntries() && val.size() != other.size())
>                         || !Objects.equals(val.updateCounter(), 
> other.updateCounter())))
>                     wrnGrps.add(part.groupId());
>             });
>         }
>         if (noData)
>             return;
>         if (!wrnGrps.isEmpty()) {
>             throw new SnapshotWarningException("Cache partitions differ for 
> cache groups " +
>                 S.toStringSortedDistinct(wrnGrps) + ". " + WRN_MSG);
>         }
> {code}
> Looks wierd.
> {code:java}
> if (noData)
>             return;
> {code}



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to