[
https://issues.apache.org/jira/browse/HBASE-4073?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13071517#comment-13071517
]
stack commented on HBASE-4073:
------------------------------
I see we have check if a region is in transition on the current regionserver
but we should also check if its already open on this server before queuing a
new open executor (I seem to recall reviewing a patch lately that added this
but can't think what it was)
> Don't open a region we already have opened!
> -------------------------------------------
>
> Key: HBASE-4073
> URL: https://issues.apache.org/jira/browse/HBASE-4073
> Project: HBase
> Issue Type: Bug
> Reporter: stack
> Priority: Critical
> Fix For: 0.92.0
>
>
> See this thread:
> http://search-hadoop.com/?q=Errors+after+major+compaction&fc_project=HBase
> In it, Eran has snippets from a log that show us being asked open a region we
> already have opened.
> We need to make sure that the root issue is addressed -- the races around
> assignment and its timeouts -- and then after that do something like a check
> if we already have region open before we queue an open (we can't return
> message to the master from down inside the regionserver event handlers).
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
