You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@phoenix.apache.org by "James Taylor (JIRA)" <ji...@apache.org> on 2016/01/24 02:32:40 UTC
[jira] [Commented] (PHOENIX-1805) Phoenix doesn't handle splits
that happen after fetching of the first batch of records for a query
[ https://issues.apache.org/jira/browse/PHOENIX-1805?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15114098#comment-15114098 ]
James Taylor commented on PHOENIX-1805:
---------------------------------------
[~samarthjain] - I think this is the issue that [~rajeshbabu] is aiming to fix in PHOENIX-1674. Perhaps you can take a look there an recommend a solutions? It seems like it's not necessarily related to local indexes?
> Phoenix doesn't handle splits that happen after fetching of the first batch of records for a query
> --------------------------------------------------------------------------------------------------
>
> Key: PHOENIX-1805
> URL: https://issues.apache.org/jira/browse/PHOENIX-1805
> Project: Phoenix
> Issue Type: Bug
> Reporter: Samarth Jain
>
> Copy pasting email response with [~jamestaylor]
> The case that is handled today:
> - client-side caches region boundaries (HBase does this in HConnection).
> - at some point a split occurs
> - some point later, a query runs. The stale region boundaries would be used for parallelization.
> - we detect this in the first batch, update the region boundaries, and rerun the failed chunk.
> However, if the split happens after the first batch is fetched, then the query fails.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)