hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "nkeywal (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-7007) [MTTR] Study assigns to see if we can make them faster still
Date Thu, 18 Oct 2012 08:08:04 GMT

    [ https://issues.apache.org/jira/browse/HBASE-7007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13478772#comment-13478772

nkeywal commented on HBASE-7007:

That what my tests were saying in June. Hopefully they're right ;-). I've not redone them
recently. Hopefully it's still there.

(from HBASE-5843)
I tested the following scenarios, on a local machine, a pseudo
distributed cluster with ZooKeeper and HBase writing in a ram drive,
no datanode nor namenode, with 2 region servers, and one empty table
with 10000 regions, 5K on each RS. Versions taken monday 2nd

3) Start of the cluster after a clean stop; wait for all regions to
become online.
0.92: ~1020s
0.94: ~1023s (tested once only)
0.96: ~31s

> [MTTR] Study assigns to see if we can make them faster still
> ------------------------------------------------------------
>                 Key: HBASE-7007
>                 URL: https://issues.apache.org/jira/browse/HBASE-7007
>             Project: HBase
>          Issue Type: Improvement
>            Reporter: stack
> Looking at a cluster start, I saw that it took about 25 minutes to assign and open 17k
regions.  8 minutes was bulk assigning via zk.  17 minutes was opening the regions.  "HBASE-6640
[0.89-fb] Allow multiple regions to be opened simultaneously" in trunk would help but maybe
we can do less work up in zk for instance; e.g. if 3.4.5, we can do bulk ops in zk (or make
a bulk assign znode that has many regions instead of one)

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

View raw message