[
https://issues.apache.org/jira/browse/PHOENIX-6702?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17550922#comment-17550922
]
Richárd Antal commented on PHOENIX-6702:
----------------------------------------
Hi
I was looking into this issue but got stock, anyway let me share my
observations, I hope it could help you at least a little.
I build HBase with and without HBASE-25709 and ran some test I think we can
safely say that it brakes the indexing, not sure how.
I looked into the failing ConcurrentMutationsExtendedIT,
PartialIndexRebuilderIT tests.
ConcurrentMutationsExtendedIT:
As [~kozdemir] mentioned we truncate the index and then after the rebuild the
index will not be correct.
I've edited scrutinizeIndex function in IndexScrutiny to not fail when the
index and table values are not the same but only at the and to be able to
compare the number of the rows.
Turns out, here we have 1 extra row.
Done some debugging and what I found interesting is that in
IndexRebuildRegionScanner.next function indexRowKeyforReadRepair is not null
and singleRowRebuildReturnCode is 1 which means no NO_INDEX_ROW
[https://github.com/apache/phoenix/blob/3d6b1206dc3687b56c315da5cdda2d7568125215/phoenix-core/src/main/java/org/apache/phoenix/coprocessor/IndexRebuildRegionScanner.java#L288]
This call does not happen when the test runs successfully.
(indexRowKeyforReadRepair is always null)
I wasn't able to debug where we set these values and why, but it is suspicious
to me.
I wasn't looking into PartialIndexRebuilderIT deeply but as far as I understand
we start the indexrebuilding and random mutations (at the same time?) and the
index ends up being wrong.
Here I changed the nThreads to 1 and the issue still appeared, for me there
were 71 rows in the index and 51 rows in the actual table, so here we should
run into the problematic code more often.
> ConcurrentMutationsExtendedIT and PartialIndexRebuilderIT fail on Hbase
> 2.4.11+
> -------------------------------------------------------------------------------
>
> Key: PHOENIX-6702
> URL: https://issues.apache.org/jira/browse/PHOENIX-6702
> Project: Phoenix
> Issue Type: Bug
> Components: core
> Affects Versions: 5.2.0, 5.1.3
> Reporter: Istvan Toth
> Assignee: Kadir OZDEMIR
> Priority: Blocker
> Fix For: 5.2.0
>
> Attachments: bisect.sh
>
>
> On my local machine
> ConcurrentMutationsExtendedIT.testConcurrentUpserts failed 6 out 10 times
> while PartialIndexRebuilderIT.testConcurrentUpsertsWithRebuild failed 10 out
> of 10 times with HBase 2.4.11 (the default build)
> The same tests succeeded 3 out of 3 times with HBase 2.3.7.
> Either HBase 2.4 has a bug, or our compatibility modules need to be fixed.
--
This message was sent by Atlassian Jira
(v8.20.7#820007)
