You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@phoenix.apache.org by "Geoffrey Jacoby (Jira)" <ji...@apache.org> on 2019/12/16 23:09:00 UTC
[jira] [Commented] (PHOENIX-5626) GlobalIndexChecker should
subclass BaseScannerRegionObserver
[ https://issues.apache.org/jira/browse/PHOENIX-5626?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16997691#comment-16997691 ]
Geoffrey Jacoby commented on PHOENIX-5626:
------------------------------------------
[~vincentpoon] [~larsh] and [~kadir], very interested in your thoughts on this one. :-)
> GlobalIndexChecker should subclass BaseScannerRegionObserver
> ------------------------------------------------------------
>
> Key: PHOENIX-5626
> URL: https://issues.apache.org/jira/browse/PHOENIX-5626
> Project: Phoenix
> Issue Type: Bug
> Reporter: Geoffrey Jacoby
> Assignee: Geoffrey Jacoby
> Priority: Major
>
> All read-path coprocessors in Phoenix (e.g ScanRO, GroupedAggregateRO and UngroupedAggregateRO) appear to subclass BaseScannerRegionObserver. This provides logic for dealing with out-of-region or split scans, and support for lookback queries in the face of deletes (see PHOENIX-4277)
> GlobalIndexChecker, which does index read repair before a subsequent region observer performs its usual logic, does not currently subclass BaseScannerRegionObserver. This seems incorrect to me, but I haven't fully worked through the implications yet. Filing a bug for discussion and possible fix.
> Thanks, [~abhishek.chouhan] for pointing me in the direction of PHOENIX-4277.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)