[Linux-HA] My resource was removed from cib.xml automatically

[Andrew Beekhof]

On 7/11/07, Stefan Lasiewski <Mailinglists at stefanco.com> wrote:
>
> I am attempting to get the resource 'nvr_hacluster' running with
> heartbeat. However, the resource never starts, and is eventually
> removed from xib.xml
>
> In this example, I want to add the resource
> 'nvr_hacluster::testsystem'. I added the resource to 'haresources'
>
> [fs1:~] cat /etc/ha.d/haresources
> fs1.c3cmsint.convio.net \
> 10.0.1.147/23/bond0 \
> nvr_hacluster::testsystem \
> nfslock nfs
>
> #### I remove cib.xml & cib.xml.sig
>
> [fs1:~] /usr/lib64/heartbeat/haresources2cib.py
> cib.xml or cib.xml.sig exist in /var/lib/heartbeat/crm/
> please remove them before generating new cib.xml
> [fs1:~] rm -f /var/lib/heartbeat/crm/cib.xml rm -f
> /var/lib/heartbeat/crm/cib.xml.sig
> [fs1:~] /usr/lib64/heartbeat/haresources2cib.py
>
>
> I can see the resource listed in cib.xml :
>
> [fs1:ha.d]# grep hacluster /var/lib/heartbeat/crm/*
> /var/lib/heartbeat/crm/cib.xml:                         <primitive
> class="heartbeat" id="nvr_hacluster_2" provider="heartbeat"
> type="nvr_hacluster">
>
> However, I never see this resource listed in 'crm_resource -L'. But I
> do see the service for 'nfs', 'nfslock' and the IPaddr:
>
> (on fs1)
> [fs1:~] crm_resource -L
> Resource Group: group_1
>     IPaddr_10_0_1_147  (heartbeat::ocf:IPaddr)
>     nfslock_2   (lsb:nfslock)
>     nfs_3       (lsb:nfs)
>
> (on fs2)
> [fs2:~] crm_resource -L
> Resource Group: group_1
>     IPaddr_10_0_1_147  (heartbeat::ocf:IPaddr)
>     nfslock_2   (lsb:nfslock)
>     nfs_3       (lsb:nfs)
>
>
> Eventually, this resource is reomved from cib.xml . There are no
> messages in the logfiles about this, so I'm not sure why it was
> removed:


It wasn't removed.

What you have just demonstrated is why modifying the the CIB on-disk is a
bad idea and why by-passing the checks we put in place to detect it is even
worse.

The other node (fs2) had a "better" version^ of the config and as soon as it
joined the cluster the entire configuration containing nvr_hacluster was
replaced with the one on fs2.

^ Version is the tupple (admin_epoche, epoche, num_updates).


[fs1:~]# grep hacluster /var/lib/heartbeat/crm/cib.xml
> [fs1:~]#
>
>
> What happened? I assume there was an error of some sort, but nothing
> is ever printed to the logs.
>
> -= Stefan
> _______________________________________________
> Linux-HA mailing list
> Linux-HA at lists.linux-ha.org
> http://lists.linux-ha.org/mailman/listinfo/linux-ha
> See also: http://linux-ha.org/ReportingProblems
>