[
https://issues.apache.org/jira/browse/PHOENIX-4033?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16091195#comment-16091195
]
Ankit Singhal commented on PHOENIX-4033:
----------------------------------------
yeah, the goal was to do the rebuild the indexes in the batches with the
checkpointing so that we don't usually time out, also give chance to other
indexes and will not lose the work done for the large rebuild when region
server holding SYSTEM.CATALOG restarted.
I think, we can't use PageFilter as we don't know from which point we should
start for the next batch.
> Limit number of rows instead of time period for partial index rebuilder
> -----------------------------------------------------------------------
>
> Key: PHOENIX-4033
> URL: https://issues.apache.org/jira/browse/PHOENIX-4033
> Project: Phoenix
> Issue Type: Bug
> Reporter: James Taylor
>
> There's a mechanism in MetaDataRegionObserver to limit the time period of the
> partial index rebuilder (based on
> QueryServices.INDEX_FAILURE_HANDLING_REBUILD_PERIOD). Assuming the goal is to
> limit the number of rows being processed during one scan, we should push a
> PageFilter instead, since limiting the time period may not limit the number
> of rows processed at a time.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
