[Pacemaker] Same host displayed twice in crm status

[David Vossel]

----- Original Message -----
> From: "Nicolas J." <nikkro70+pacemaker at gmail.com>
> To: "The Pacemaker cluster resource manager" <pacemaker at oss.clusterlabs.org>
> Sent: Tuesday, April 2, 2013 2:07:14 AM
> Subject: Re: [Pacemaker] Same host displayed twice in crm status
> 
> I've already tried to remove the node following the document:
> http://clusterlabs.org/doc/en-US/Pacemaker/1.0/html/Pacemaker_Explained/s-node-delete.html
> 
> with the following commands:
> crm_node -R COROSYNC_ID
> cibadmin --delete --obj_type nodes --crm_xml '<node uname="
> VMTESTORADG2.it.dbi-services.com "/>'
> cibadmin --delete --obj_type status --crm_xml '<node_state uname="
> VMTESTORADG2.it.dbi-services.com "/>'
> 
> I also tried to delete the status before deleting the node with the same
> results.
> 
> I have the same issue, the node is deleted but when corosync is restarted on
> any node, the deleted name appears again.
> The server with the changed hostname has been restarted so I don't know where
> the reference to the old name can be saved except in the cluster.

Unfortunately this looks like a bug.  It sounds like the crm_node -R option isn't properly discarding the node cache, which means every time a new policy engine transition is generated that node slips back in.  I'm not sure what to tell you. I'm guessing the only way to fix this is to shutdown the cluster entirely making sure their is on mention of the old node in the config on startup, or trying a new version of pacemaker.  I'm sure neither of those solutions are what you want to hear though.  Maybe someone else has some better advice who has encountered this with your version.

-- Vossel

> Regarding the version, here are the details:
> - Corosync 1.2.7-1.1.el5
> - Pacemaker 1.1.5-1.1.el5
> 
> 
> 2013/4/1 David Vossel < dvossel at redhat.com >
> 
> 
> 
> ----- Original Message -----
> > From: "Nicolas J." < nikkro70+pacemaker at gmail.com >
> > To: pacemaker at oss.clusterlabs.org
> > Sent: Friday, March 29, 2013 8:55:30 AM
> > Subject: [Pacemaker] Same host displayed twice in crm status
> > 
> > Hi,
> > 
> > I have a problem with a Corosync/Pacemaker configuration.
> > One host of the cluster has been renamed and now the host is displayed
> > twice
> > in the configuration.
> > 
> > When I try to remove the host from the configuration it works but if
> > corosync
> > is restarted on one node, the old host appears again.
> > I tried several ways to delete the host with no effect.
> > 
> > How can I delete the wrong host?
> 
> For the pacemaker version you are using, try deleting the node from the
> configuration in both the <node> and <status> sections, then use crm_node -R
> option to remove the node from the cluster's internal cache. In pacemaker
> versions >= 1.1.8 only the crm_node -R option is required to remove a node.
> 
> -- Vossel
> 
> > I checked the Linux configuration and there is no place where the old name
> > is
> > referenced.
> > It's an OEL/Red Hat linux.
> > 
> > Output
> > ---------
> > [root at vmtestoradg2 ~]# crm status
> > ============
> > Last updated: Fri Mar 29 14:51:56 2013
> > Stack: openais
> > Current DC: vmtestoradg1 - partition with quorum
> > Version: 1.1.5-1.1.el5-01e86afaaa6d4a8c4836f68df80ababd6ca3902f
> > 4 Nodes configured, 3 expected votes
> > 1 Resources configured.
> > ============
> > 
> > Online: [ vmtestoradg1 vmtestora10g01 vmtestoradg2 ]
> > OFFLINE: [ VMTESTORADG2.it.dbi-services.com ]
> > 
> > DG_IP (ocf::heartbeat:IPaddr2): Started vmtestoradg1
> > 
> > [root at vmtestoradg2 ~]# crm node clearstate VMTESTORADG2.it.dbi-services.com
> > Do you really want to drop state for node VMTESTORADG2.it.dbi-services.com
> > ?
> > y
> > [root at vmtestoradg2 ~]# crm node delete VMTESTORADG2.it.dbi-services.com
> > INFO: node VMTESTORADG2.it.dbi-services.com not found by crm_node
> > INFO: node VMTESTORADG2.it.dbi-services.com deleted
> > 
> > Thanks in advance
> > 
> > Best Regards,
> > 
> > Nicolas J.
> > 
> > _______________________________________________
> > 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
> > 
> 
> _______________________________________________
> 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
> 
> 
> _______________________________________________
> 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
>