cassandra-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Alex Araujo (Commented) (JIRA)" <>
Subject [jira] [Commented] (CASSANDRA-2967) Only bind JMX to the same IP address that is being used in Cassandra
Date Fri, 18 Nov 2011 04:27:52 GMT


Alex Araujo commented on CASSANDRA-2967:

Vijay, you are correct - glad you chimed in.  I ran the patch with {{-Dsun.rmi.dgc.server.gcInterval=60000}}
and this was the output:

2011-11-17T21:06:03.832-0600: 2.809: [Full GC 29495K->1075K(1028096K), 0.0484004 secs]
2011-11-17T21:07:03.919-0600: 62.894: [Full GC 73330K->2885K(1028096K), 0.0668785 secs]
2011-11-17T21:08:03.993-0600: 122.967: [Full GC 16014K->2824K(1028096K), 0.0584933 secs]
2011-11-17T21:09:04.056-0600: 183.029: [Full GC 15953K->2823K(1028096K), 0.0548553 secs]
2011-11-17T21:10:04.118-0600: 243.089: [Full GC 17583K->2823K(1028096K), 0.0513944 secs]
2011-11-17T21:11:04.175-0600: 303.145: [Full GC 24671K->2858K(1028096K), 0.0547190 secs]
2011-11-17T21:12:04.236-0600: 363.205: [Full GC 21430K->2826K(1028096K), 0.0535503 secs]
2011-11-17T21:13:04.295-0600: 423.263: [Full GC 17574K->2823K(1028096K), 0.0539739 secs]
2011-11-17T21:14:04.355-0600: 483.322: [Full GC 17589K->2822K(1028096K), 0.0500845 secs]
2011-11-17T21:15:04.412-0600: 543.377: [Full GC 19214K->2822K(1028096K), 0.0578777 secs]

{{-XX:-DisableExplicitGC}} does not disable Full GC's (only explicit {{System.gc}} calls as
the name implies), and increasing gcInterval does seem a bit dangerous.  At best, we can enable
the code in the patch if the values are uncommented in cassandra.yaml, but not sure that's
any more useful than the current options.
> Only bind JMX to the same IP address that is being used in Cassandra
> --------------------------------------------------------------------
>                 Key: CASSANDRA-2967
>                 URL:
>             Project: Cassandra
>          Issue Type: Bug
>          Components: Tools
>    Affects Versions: 0.8.2
>            Reporter: Joaquin Casares
>            Assignee: Alex Araujo
>            Priority: Minor
>              Labels: lhf
>         Attachments: cassandra-0.8-2967.txt, cassandra-1.0-2967-v2.txt, cassandra-1.0-2967-v3.txt,
> The setup is 5 nodes in each data center are all running on one physical test machine
and even though the repair was run against the correct IP the wrong JMX port was used. As
a result, instead of repairing all 5 nodes I was repairing the same node 5 times.
> It would be nice if Cassandra's JMX would bind to only the IP address on which its thrift/RPC
services are listening on instead of binding to all IP's on the box.

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:!default.jspa
For more information on JIRA, see:


View raw message