[Pacemaker] pingd CPU usage increases slowly

Andrew Beekhof andrew at beekhof.net
Tue May 12 18:22:16 UTC 2009


Excellent news.
The slowdown was probably related to the memory leak I fixed for 1.0.3

Let me know if you have any further problems

On Tue, May 12, 2009 at 7:42 PM, Stelio Plautz <stelio at sphome.net> wrote:
> Am Tue, 12 May 2009 11:19:46 +0200
> schrieb Andrew Beekhof <andrew at beekhof.net>:
>
> Hi Andrew,
> I've upgraded both nodes to 1.0.3 and it looks ok now.
>
> Thanks stelio
>
>
>> On Mon, May 11, 2009 at 12:01 PM, Stelio Plautz <stelio at sphome.net>
>> wrote:
>> > Hi all,
>> >
>> > I've set up a 2 node cluster on debian etch amd64, pacemaker 1.0.2-1
>> > and heartbeat 2.99.1-1 from suse repository.
>> > everything works fine, but pingd increases CPU usage slowly. I've
>> > two pingd processes running and both use about 100 % CPU after 3
>> > weeks.
>> >
>> > 14091 root      16   0 1299M 1263M   936 S 55.6  7.4
>> > 166h44:20 /usr/lib/heartbeat/pingd -D -p /var/run/public_net.pid -a
>> > public_net -d 5s -m 100 -h xxx.xxx.xxx.1 14098 root      15   0
>> > 1299M 1263M   936 R 47.0  7.4 166h42:22 /usr/lib/heartbeat/pingd -D
>> > -p /var/run/mgmt_net.pid -a mgmt_net -d 5s -m 100 -h zzz.zzz.zzz.1
>> >
>> > I've found the following link ...
>> > http://www.gossamer-threads.com/lists/linuxha/pacemaker/54236
>> > ... and installed the pingd RA from pacemaker 1.0.3-1.
>> > The default ping interval is now 1, but pingd CPU usage still
>> > slow-growing.
>> >
>> > Thanks for any advice
>>
>> I'm pretty sure this is fixed in 1.0.3 (you'd need the new binary as
>> well)
>>
>> _______________________________________________
>> Pacemaker mailing list
>> Pacemaker at oss.clusterlabs.org
>> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>
>
>
> --
> lg stelio
>
> _______________________________________________
> Pacemaker mailing list
> Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>




More information about the Pacemaker mailing list