[Pacemaker] trouble with crm configure load replace / crm config in a VCS

Dejan Muhamedagic dejanmm at fastmail.fm
Thu Feb 4 09:53:11 UTC 2010


Hi,

On Thu, Feb 04, 2010 at 10:03:58AM +0100, Marc Fournier wrote:
> Hello,
> 
> I'm quite new to pacemaker and I would like to have my cluster
> configuration revisioned in a VCS.
> 
> My idea is to edit the configuration in a text file, then load it onto a
> test cluster. Once I'm happy with my changes, push the file to one of the
> production machines via the VCS and load it into the production cluster.
> 
> Unfortunately I'm having some issues with "crm configure load replace".
> This command seems to stop, then remove all resources before re-creating
> and starting them again (maybe I'm wrong).

You're not wrong. Before applying the new configuration, the old
one is erased. The CRM immediately acts on the new configuration.

> As show below, just dumping the current configuration to a file, and
> loading it again, fails.
> 
> I wanted to use "crm_diff" and "cibadmin --patch" instead, but I was unable
> to find a way to convert crm resources to xml without passing them through
> a running cluster (crm configure load + crm configure save xml).
> 
> So my questions are:
> 
>  - should "crm configure load replace" work as I naively expect ? Maybe I
>    just have something wrong in my configuration ?

It should work that way.

>  - how can I convert a file containing crm cli commands to an xml cib file ?

configure save xml

>  - any suggestions on the best way to have my cluster configuration
>    revisioned in a VCS ?

Can you please open a bugzilla and post hb_report. I am a bit
mystified by what you shown below.

Thanks,

Dejan

> Thanks !
> Marc
> 
> 
> ----8<-------------------------------------------------------------------
> 
> crm(live)# status
> 
> 
> ============
> Last updated: Thu Feb  4 09:22:50 2010
> Stack: Heartbeat
> Current DC: bob (f407f405-e4ec-47b2-8865-b22c322d208e) -
> partition with quorum 2 Nodes configured, unknown expected votes
> 5 Resources configured.
> ============
> 
> Online: [ alice bob ]
> 
> Resource Group: mysql
>     fs_mysql    (ocf::heartbeat:Filesystem):    Started bob
>     svc_mysql   (lsb:mysqld):   Started bob
> Resource Group: tomcat
>     fs_lucene   (ocf::heartbeat:Filesystem):    Started bob
>     svc_tomcat  (lsb:tomcat-gedscr):    Started bob
> Master/Slave Set: ms_drbd_lucene
>         Masters: [ bob ]
>         Slaves: [ alice ]
> Master/Slave Set: ms_drbd_mysql
>         Masters: [ bob ]
>         Slaves: [ alice ]
> Resource Group: frontend
>     svc_apache  (ocf::heartbeat:apache):        Started bob
>     ip_gedscr   (ocf::heartbeat:IPaddr2):       Started bob
> 
> 
> crm(live)# configure save /tmp/test.cli
> crm(live)# configure load replace /tmp/test.cli
> 
> crm(live)# status 
> 
> 
> ============
> Last updated: Thu Feb  4 09:37:30 2010
> Stack: Heartbeat
> Current DC: bob (f407f405-e4ec-47b2-8865-b22c322d208e) - partition with quorum
> 2 Nodes configured, unknown expected votes
> 5 Resources configured.
> ============
> 
> Online: [ alice bob ]
> 
> Master/Slave Set: ms_drbd_mysql
>     drbd_mysql:1        (ocf::linbit:drbd):     Slave bob (unmanaged) FAILED
>         Stopped: [ drbd_mysql:0 ]
> 
> Failed actions:
>     drbd_mysql:1_demote_0 (node=bob, call=300, rc=6, status=complete): not configured
>     drbd_mysql:1_stop_0 (node=bob, call=303, rc=6, status=complete): not configured
>     drbd_lucene:1_demote_0 (node=bob, call=304, rc=6, status=complete): not configured
> 
> 
> 
> 
> _______________________________________________
> Pacemaker mailing list
> Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker




More information about the Pacemaker mailing list