[
https://issues.apache.org/jira/browse/HBASE-5569?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13232359#comment-13232359
]
Lars Hofhansl commented on HBASE-5569:
--------------------------------------
This
{code}
if (includeDeleteMarker
&& kv.getMemstoreTS() <= maxReadPointToTrackVersions) {
this.deletes.add(bytes, offset, qualLength, timestamp, type);
}
{code}
Fixes the issue. Note that maxReadPointToTrackVersions is actually the minimum
readpoint of any scanner still operating in the region and it is *only* set
during compaction.
I think this correct because of the following:
All delete markers precede the KVs they affect. So by not adding the delete
marker it is guarantees that no KVs will be removed during flush that might
still be in use. It also removes this race condition between scanner and
flushes.
So my previous fix was almost correct (in thought at least). I had believed it
to be correct, because I had not been able - not even a single time - to
reproduce this on my work machine.
I'll attach a patch soon.
> Do not collect deleted KVs when they are still in use by a scanner.
> -------------------------------------------------------------------
>
> Key: HBASE-5569
> URL: https://issues.apache.org/jira/browse/HBASE-5569
> Project: HBase
> Issue Type: Bug
> Reporter: Lars Hofhansl
> Assignee: Lars Hofhansl
> Fix For: 0.94.0, 0.96.0
>
> Attachments: 5569-v2.txt, 5569.txt,
> TestAtomicOperation-output.trunk_120313.rar
>
>
> I noticed this because TestAtomicOperation.testMultiRowMutationMultiThreads
> fails rarely.
> The solution is similar to HBASE-2856, where expired KVs are not collected
> when in use by a scanner.
> ---
> What I pieced together so far is that it is the *scanning* side that has
> problems sometimes.
> Every time I see a assertion failure in the log I see this before:
> {quote}
> 2012-03-12 21:48:49,523 DEBUG [Thread-211] regionserver.StoreScanner(499):
> Storescanner.peek() is changed where before =
> rowB/colfamily11:qual1/75366/Put/vlen=6,and after =
> rowB/colfamily11:qual1/75203/DeleteColumn/vlen=0
> {quote}
> The order of if the Put and Delete is sometimes reversed.
> The test threads should always see exactly one KV, if the "before" was the
> Put the thread see 0 KVs, if the "before" was the Delete the threads see 2
> KVs.
> This debug message comes from StoreScanner to checkReseek. It seems we still
> some consistency issue with scanning sometimes :(
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
