[
https://issues.apache.org/jira/browse/HBASE-17625?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jimmy Xiang updated HBASE-17625:
--------------------------------
Description:
Tried to enable a table with 10k+ regions, it takes more time to generate the
plan than do the actual assignment. This is so embarrassing. :)
It turns out that it took quite some time to get the top HDFS block locations
in registering regions when creating the Cluster object.
was:
Tried to enable a table with 10k+ regions, it takes more time to generate the
plan than do the actual assignment. This is so embarrassing. :)
It turns out that it took quite some time to get the top HDFS block locations
in registering regions when creating the Cluster object. There is no new region
server, why do we need such info when trying to retain assignment?
Is the region availability thing related to region replica? Can we avoid such
penalty if region replica in not needed?
> Slow to enable a table
> ----------------------
>
> Key: HBASE-17625
> URL: https://issues.apache.org/jira/browse/HBASE-17625
> Project: HBase
> Issue Type: Improvement
> Reporter: Jimmy Xiang
>
> Tried to enable a table with 10k+ regions, it takes more time to generate the
> plan than do the actual assignment. This is so embarrassing. :)
> It turns out that it took quite some time to get the top HDFS block locations
> in registering regions when creating the Cluster object.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
