[Pacemaker] Possible bug?

[Andrew Beekhof]

On Tue, Sep 11, 2012 at 4:58 PM, Borislav Borisov
<borislav.v.borisov at gmail.com> wrote:
>
>> Pacemaker creates a second cib process to write the contents to disk
>> after a change so that the 'real' process doesn't block.
>>
>> After the process writes the cib to disk, we then try to read it back
>> again to verify that everything is sane.
>> What you're seeing here is that check failing for some reason.
>
>
> Because this much I could tell from the code, I was perplexed from that
> behavior.

Would you not say its better to know /now/ that what we wrote to disk
is unusable (or just not what you intended)?
Or would you prefer to find out the next time the node booted :)

Either way, Pacemaker will continue to function normally - the current
CIB just wont be written to disk on that one node.

>>
>> Looking at the logs I see:
>>
>> Sep 10 15:24:40 Cluster-Server-1 cib: [54564]: ERROR:
>> validate_cib_digest: Digest comparision failed: expected
>> 83751b899e758f9b138d060ace084080 (/var/lib/heartbeat/crm/cib.ANRY1Q),
>> calculated bd97ef3df10846e783bd64059be77e45
>> Sep 10 15:24:40 Cluster-Server-1 cib: [54564]: ERROR: retrieveCib:
>> Checksum of /var/lib/heartbeat/crm/cib.uGGnOm failed!  Configuration
>> contents ignored!
>>
>> Which is really strange.
>> I see a couple of changes in related areas since 1.1.7, perhaps one of
>> those will fix your issue.
>> (1.1.8 should be out today/tomorrow)
>
>
> I guess that I will wait for the next release and see if the issue is fixed
> in it.
>
> Anyway, thanks a bunch Andrew.
>
>
> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
>
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://bugs.clusterlabs.org
>