[Pacemaker] pingd fails to update CIB

Andrew Beekhof andrew at beekhof.net
Mon Mar 15 11:00:03 UTC 2010


On Sat, Mar 13, 2010 at 1:26 AM, Quentin Smith <quentin at mit.edu> wrote:
> I don't know a lot about hg, but doesn't the "r15404" in the version of
> pacemaker that I'm running now mean that I already have this bugfix
> (r15295)?

Yes.
I'd suggest you use the ping RA instead of pingd.

The ping RA uses your system's ping binary which is more robust than
pingd has proven to be :-(

>
> --Quentin
>
> On Fri, 12 Mar 2010, hj lee wrote:
>
>> Hi,
>>
>> This seems the same problem I reported a while ago. It was fixed in
>> http://hg.clusterlabs.org/pacemaker/stable-1.0/rev/214f0fc258f2.
>>
>> Thanks
>>
>> On Fri, Mar 12, 2010 at 2:36 PM, Quentin Smith <quentin at mit.edu> wrote:
>>      Hi-
>>
>>      I just took the latest updates to pacemaker and heartbeat from
>> http://people.debian.org/~madkiss/ha. In particular, I upgraded
>>
>>      heartbeat 1:3.0.2-1~bpo50+1 to 1:3.0.2+hg12547-2~bpo50+1
>>      pacemaker 1.0.7+hg20100203-1~bpo50+1 to
>> 1.0.7+hg20100303r15404-3~bpo50+1
>>      cluster-agents 1:1.0.2-1~bpo50+1 to 1:1.0.2+hg1657-2~bpo50+1
>>      cluster-glue 1.0.3-1~bpo50+1 to 1.0.3+hg2337-1~bpo50+1
>>
>>      Now, my pingd resources don't appear to update the node attributes
>> when a ping node reappears. In particular, if I block the target of a
>>      pingd, the node's attribute is set to 0, but if I unblock it, the
>> node's attribute is not reset in the CIB. Running "attrd_updater -R" causes
>>      attrd to update the CIB, and the node's attribute goes back to 1000
>> (my chosen multiplier). I tried restarting attrd, and that didn't help.
>>      When pingd/attrd should be updating the attribute, instead it logs
>> "info: ping_read: Retrying...".
>>
>>      I spoke to misch on IRC and he suggested mailing the list. My cluster
>> is quorate, and other resources run fine. The pingd primitive is:
>>
>>      primitive pingd_w20net ocf:pacemaker:pingd \
>>       params name="pingd_w20net" dampen="5s" multiplier="1000"
>> host_list="18.187.0.1"
>>
>>      I'd appreciate any advice you have for bringing my cluster back to
>> full operation.
>>
>>      --Quentin
>>
>>      _______________________________________________
>>      Pacemaker mailing list
>>      Pacemaker at oss.clusterlabs.org
>>      http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>>
>>
>>
>>
>> --
>> Peakpoint Service
>>
>> Cluster Setup, Troubleshooting & Development
>> kerdosa at gmail.com
>> (303) 997-2823
>>
>
> _______________________________________________
> Pacemaker mailing list
> Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>
>




More information about the Pacemaker mailing list