You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@hbase.apache.org by "Daisuke Kobayashi (JIRA)" <ji...@apache.org> on 2015/09/11 08:29:45 UTC

[jira] [Created] (HBASE-14408) Separate RPC pool for metadata and data related RPCs, for better HBCK use

Daisuke Kobayashi created HBASE-14408:
-----------------------------------------

             Summary: Separate RPC pool for metadata and data related RPCs, for better HBCK use
                 Key: HBASE-14408
                 URL: https://issues.apache.org/jira/browse/HBASE-14408
             Project: HBase
          Issue Type: Improvement
          Components: hbck, rpc
    Affects Versions: 1.0.0
            Reporter: Daisuke Kobayashi


While a regionserver is in busy state, such like plenty of responseTooSlow being thrown, hbck fails because it cannot reach out to the particular regionserver due to SocketTimeoutException. In such scenario, bunch of inconsistencies could be detected, while the regionserver itself doesn't go down, nor does inconsistencies exist actually.

That should lead a lot of users off track and into a -repair frenzy.

I'd be nice to have a separate RPC pool for such requests from Hbck.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)