[
https://issues.apache.org/jira/browse/HBASE-11839?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Sergey Soldatov updated HBASE-11839:
------------------------------------
Attachment: HBASE-11839-v1.patch
Quite simple patch. The problem is that average load calculated using both
online and dead servers. An example:
25 regions on 3 region servers with 9, 9, 7 regions correspondingly. 3rd is
going down and its regions randomly copied to other regions, so, for example,
it become as 15,10 and 3rd server which is down is still having 7. Current code
calculated average as (15 + 10 + 7)/3 = 10.6 and this value is used during
computation instead of 12.5 (25 regions and 2 online servers).
Patch uses only online servers during average computation.
Please correct me if I wrong.
> TestRegionRebalance is flakey
> -----------------------------
>
> Key: HBASE-11839
> URL: https://issues.apache.org/jira/browse/HBASE-11839
> Project: HBase
> Issue Type: Bug
> Reporter: Alex Newman
> Assignee: Alex Newman
> Fix For: 2.0.0, 0.99.1
>
> Attachments: HBASE-11839-v1.patch
>
>
> Besides failing many times on the prebuild TestRegionRebalance fails on my
> local machine eventually simply with
> export RUNNIN=true; mvn clean install -DskipTests ; while ($RUNNIN) ; do mvn
> test -Dtest=TestRegionRebalancing || RUNNIN=false;done
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
