[
https://issues.apache.org/jira/browse/HBASE-55?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12570058#action_12570058
]
Jim Kellerman commented on HBASE-55:
------------------------------------
Right now the master gives the region server one
hbase.hbasemaster.maxregionopen (currently the same as
hbase.master.lease.period) to acknowledge that it received the open request
with a HMsg.MSG_REPORT_PROCESS_OPEN If the master receives that message, it
adds an additional maxregionopen time period for the region server to send a
HMsg.MSG_REPORT_OPEN (the region is now on-line), before the master will
reassign the region.
It is possible that it could take longer than that amount of time if there are
a lot of log entries to apply or if the initial compaction on region open takes
a while or if we get into one of those situations in which the region server
cannot talk to the master (I'd really like to find out why this happens).
Currently the region server only reports in once per hbase.master.lease.period.
This is probably too infrequent as we often see both MSG_REPORT_PROCESS_OPEN
and MSG_REPORT_OPEN in the same set of messages.
Probably maxregionopen should be greater than lease.period or we should add a
new parameter "heartbeat.interval" which would be less than the lease period so
the region server would report in more frequently.
And, if the master receives a MSG_REPORT_PROCESS_OPEN, it should not take any
action with respect to reassigning the region unless the region server's lease
times out.
> [hbase] Improve Master region assignment function
> -------------------------------------------------
>
> Key: HBASE-55
> URL: https://issues.apache.org/jira/browse/HBASE-55
> Project: Hadoop HBase
> Issue Type: Improvement
> Components: master
> Reporter: Bryan Duxbury
> Fix For: 0.2.0
>
>
> We would like the master's region assignment function to take into account
> more factors when choosing where to assign regions.
>
> - More advanced accounting of load on regionserver - memory, # requests, etc
> - Don't deploy both daughter regions to the same regionserver
> - Assign regions where the underlying DFS blocks are hosted if possible
> Please add additional ideas in comments as they come up.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
