You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ambari.apache.org by Jaimin Jetly <ja...@hortonworks.com> on 2016/02/29 23:24:17 UTC
Review Request 44183: Display tag for StandBy name node should not be
changed to NameNode if there is an active namenode
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/44183/
-----------------------------------------------------------
Review request for Ambari, Srimanth Gunturi and Yusaku Sako.
Bugs: AMBARI-15235
https://issues.apache.org/jira/browse/AMBARI-15235
Repository: ambari
Description
-------
On a large cluster with heavy load on NameNode, sometimes there happens to be greater latency for Standby NameNode in responding back its JMX metrics which results in timeout for the JMX call dome by ambari-server and no JMX data for standby NameNode is received sometimes.
Thus Standby NameNode in such environment responses sometimes by declaring itself standby and sometimes doesn't respond quick enough before timeout happens.
Due to this behavior ambari-web keeps alternating frequently between “standby namenode” and “namenode”, even though there is no failover.
As a fix ambari if detects any one namenode to be active should consider other to be standby even though JMX call to standby namenode has timedout.
Diffs
-----
ambari-web/app/mappers/service_metrics_mapper.js a88c3e6
ambari-web/test/mappers/service_metrics_mapper_test.js d2c9cc0
Diff: https://reviews.apache.org/r/44183/diff/
Testing
-------
Tested the work manually on a NameNode HA cluster.
Verified that all ambari-web unit test passes successfully with the patch:
24467 tests complete (24 seconds)
146 tests pending
Thanks,
Jaimin Jetly
Re: Review Request 44183: Display tag for StandBy name node should
not be changed to NameNode if there is an active namenode
Posted by Yusaku Sako <yu...@hortonworks.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/44183/#review121335
-----------------------------------------------------------
Ship it!
Ship It!
- Yusaku Sako
On Feb. 29, 2016, 10:24 p.m., Jaimin Jetly wrote:
>
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/44183/
> -----------------------------------------------------------
>
> (Updated Feb. 29, 2016, 10:24 p.m.)
>
>
> Review request for Ambari, Srimanth Gunturi and Yusaku Sako.
>
>
> Bugs: AMBARI-15235
> https://issues.apache.org/jira/browse/AMBARI-15235
>
>
> Repository: ambari
>
>
> Description
> -------
>
> On a large cluster with heavy load on NameNode, sometimes there happens to be greater latency for Standby NameNode in responding back its JMX metrics which results in timeout for the JMX call dome by ambari-server and no JMX data for standby NameNode is received sometimes.
> Thus Standby NameNode in such environment responses sometimes by declaring itself standby and sometimes doesn't respond quick enough before timeout happens.
>
> Due to this behavior ambari-web keeps alternating frequently between “standby namenode” and “namenode”, even though there is no failover.
>
> As a fix ambari if detects any one namenode to be active should consider other to be standby even though JMX call to standby namenode has timedout.
>
>
> Diffs
> -----
>
> ambari-web/app/mappers/service_metrics_mapper.js a88c3e6
> ambari-web/test/mappers/service_metrics_mapper_test.js d2c9cc0
>
> Diff: https://reviews.apache.org/r/44183/diff/
>
>
> Testing
> -------
>
> Tested the work manually on a NameNode HA cluster.
> Verified that all ambari-web unit test passes successfully with the patch:
>
>
> 24467 tests complete (24 seconds)
> 146 tests pending
>
>
> Thanks,
>
> Jaimin Jetly
>
>