[
https://issues.apache.org/jira/browse/HBASE-5063?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13172045#comment-13172045
]
Lars Hofhansl commented on HBASE-5063:
--------------------------------------
Looks like this.masterAddressManager.getMasterAddress() could return null (see
first loop), so this could lead to an NPE.
I am wondering we shouldn't just fold the check from the first loop (where we
get masterServerName) into the 2nd loop and completely remove the first loop.
I.e. if masterServerName is null, continue the loop, sleep for a bit... Means
that the sleep needs to be pulled out of the try/catch. If masterServerName is
not null, try to connect.
> RegionServers fail to report to backup HMaster after primary goes down.
> -----------------------------------------------------------------------
>
> Key: HBASE-5063
> URL: https://issues.apache.org/jira/browse/HBASE-5063
> Project: HBase
> Issue Type: Bug
> Affects Versions: 0.92.0
> Reporter: Jonathan Hsieh
> Assignee: Jonathan Hsieh
> Priority: Critical
> Attachments: HBASE-5063.patch
>
>
> # Setup cluster with two HMasters
> # Observe that HM1 is up and that all RS's are in the RegionServer list on
> web page.
> # Kill (not even -9) the active HMaster
> # Wait for ZK to time out (default 3 minutes).
> # Observe that HM2 is now active. Tables may show up but RegionServers never
> report on web page. Existing connections are fine. New connections cannot
> find regionservers.
> Note:
> * If we replace a new HM1 in the same place and kill HM2, the cluster
> functions normally again after recovery. This sees to indicate that
> regionservers are stuck trying to talk to the old HM1.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
