[
https://issues.apache.org/jira/browse/PHOENIX-5640?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16999804#comment-16999804
]
Geoffrey Jacoby commented on PHOENIX-5640:
------------------------------------------
+1.
Interesting that the long is written as a normal HBase increment that uses
Bytes.toLong internally rather than using the Phoenix codecs, but it's
consistent and always accessed through HBase APIs on both read and write, so
that's OK.
> Pending disable count should not be increased for rebuild write failures
> ------------------------------------------------------------------------
>
> Key: PHOENIX-5640
> URL: https://issues.apache.org/jira/browse/PHOENIX-5640
> Project: Phoenix
> Issue Type: Bug
> Affects Versions: 5.0.0, 4.14.3
> Reporter: Kadir OZDEMIR
> Assignee: Kadir OZDEMIR
> Priority: Major
> Attachments: PHOENIX-5640.4.x-HBase-1.5.001.patch
>
>
> For index rebuild write failures, PhoenixIndexFailurePolicy does not change
> the index state but unnecessarily increases the pending disable count that is
> maintained in the system catalog. This leads to system catalog corruption
> when view indexes experience index rebuild write failures. This is because
> PhoenixIndexFailurePolicy uses the physical table name instead of the logical
> table name when updating the system catalog.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
