[
https://issues.apache.org/jira/browse/HBASE-13316?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14378099#comment-14378099
]
Sameet Agarwal commented on HBASE-13316:
----------------------------------------
Elliott:
- If we do the Warmup asynchronously, then we will need a way to wait in master
for the warmup to be done. Otherwise we lose the benefit of the warmup since
the open will end up happening sooner than the warmup can get done. Is there a
convenient way to wait for this to happen in master or to send an asynchronous
response back from the region server once warmup is done without blocking the
thread?
- warmup and open happening concurrently should work. warmup doesn't add itself
to regions in transition. So, a new open will simply proceed as normal and open
its own files and do its start up. It may end up finding the data in cache or
it may read it off disk. I have tried the moves concurrently with other
activities happening but I dont have a test to actually run the warmup in
parallel with open
> Reduce the downtime on planned moves of regions
> -----------------------------------------------
>
> Key: HBASE-13316
> URL: https://issues.apache.org/jira/browse/HBASE-13316
> Project: HBase
> Issue Type: Improvement
> Components: Balancer
> Affects Versions: 2.0.0, 1.0.1, 1.1.1
> Reporter: Sameet Agarwal
> Assignee: Sameet Agarwal
> Priority: Minor
> Fix For: 2.0.0
>
> Attachments: PlannedFailover.patch
>
> Original Estimate: 168h
> Remaining Estimate: 168h
>
> The current behavior of a region move shuts down a region and then starts is
> up in another regionserver. This causes increased latency and possibly
> timeouts till the new region's cache is fully warmed up. We can make a region
> move less disruptive by warming the cache in the destination region server
> before shutting dow the old region.
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
