[Pacemaker] CentOS 6.4 and CFS.

Rob Thomas xrobau at gmail.com
Fri Nov 15 20:04:21 UTC 2013


Well. This is interesting.

>> Looking through the pcs code, it's now checking that the node exists
>> in /etc/corosync/corosync.conf
>
> Noooo.  Not on RHEL-6 anyway.

Line 363 of /usr/lib/python2.6/site-packages/pcs/cluster.py has this:

    nodes = utils.getNodesFromCorosyncConf()

getNodesFromCorosyncConf()  looks for 'ring0_addr:' in
settings.corosync_conf_file, which is /etc/corosync/corosync.conf

[root at localhost pcs]# rpm -qf cluster.py
pcs-0.9.90-1.0.1.el6.centos.noarch
[root at localhost pcs]# md5sum cluster.py
b22cfb3551e0c0e92fb90d287932e64d  cluster.py

I grabbed http://ftp.redhat.com/redhat/linux/enterprise/6Server/en/os/SRPMS/pcs-0.9.90-1.el6_4.src.rpm
to see if maybe CentOS had done something wrong, but the same things
are still in there.

So. Everything documented in this:

>    http://clusterlabs.org/quickstart-redhat.html

works, because that's not using anything that uses
getNodesFromCorosyncConf() which is everywhere through cluster.py

But after doing the commands in there, if you try to set a node
offline, it doesn't work, and comes up with that error that the node
doesn't exist.

--Rob




More information about the Pacemaker mailing list