[
https://issues.apache.org/jira/browse/PHOENIX-6462?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17340399#comment-17340399
]
Tanuj Khurana commented on PHOENIX-6462:
----------------------------------------
[~giskender] I was discussing this issue with [~kadir] and he brought up a
point of distinguishing between hbase client retries within a single map task
and overall map task retries. There is some logic in IndexRebuildRegionScanner
([last
update|https://github.com/apache/phoenix/blob/4.x/phoenix-core/src/main/java/org/apache/phoenix/coprocessor/IndexRebuildRegionScanner.java#L375-L382])
to resume the scan to skip rows that have been built when the hbase client
retries in cases when the region moves, splits.
> Index build mapper that failed should not be logging into the
> PHOENIX_INDEX_TOOL_RESULT table
> ---------------------------------------------------------------------------------------------
>
> Key: PHOENIX-6462
> URL: https://issues.apache.org/jira/browse/PHOENIX-6462
> Project: Phoenix
> Issue Type: Bug
> Reporter: Gokcen Iskender
> Priority: Major
>
> Today, if a mapper fails it still logs the region into the
> PHOENIX_INDEX_TOOL_RESULT table. This causes into mistakenly thinking that
> the mapper succeeded. Incremental rebuilds will be mislead.
>
> [~swaroopa] [~tkhurana]
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
