[
https://issues.apache.org/jira/browse/HBASE-20002?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16366528#comment-16366528
]
Duo Zhang commented on HBASE-20002:
-----------------------------------
I think with HBASE-20000, the problem is partially solved. We have a stuck
checker which will start new worker thread if half of the workers are stuck,
and then with HBASE-20000, first it will select ServerCrashProcedure if any,
and it will not block, and then finally it will select RecoverMetaProcedure
since it is always placed at beginning, and then we can make progress.
Then let me address the keep alive problem first, I've already filed an issue I
think. The patch would be the last choice if all other approach are failed.
> Reserve worker thread in ProcedureExecutor for high priority procedures
> -----------------------------------------------------------------------
>
> Key: HBASE-20002
> URL: https://issues.apache.org/jira/browse/HBASE-20002
> Project: HBase
> Issue Type: Sub-task
> Components: proc-v2
> Reporter: Duo Zhang
> Assignee: Duo Zhang
> Priority: Major
> Fix For: 2.0.0-beta-2
>
> Attachments: HBASE-20002.patch
>
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
