1. sniplet of cluster.conf file from one cluster :-
<?xml version="1.0"?>
<cluster alias="Cluster" config_version="1" name="Cluster">
<clusternodes>
<clusternode name="blade4-2" nodeid="2" votes="1">
<fence>
<method name="1">
<device blade="2" missing_as_off="1" name="BladeCenterFencing"/>
</method>
</fence>
</clusternode>
<clusternode name="blade4-1" nodeid="1" votes="1">
<fence>
<method name="1">
<device blade="1" missing_as_off="1" name="BladeCenterFencing"/>
</method>
</fence>
</clusternode>
</clusternodes>
<cman expected_votes="1" two_node="1"/>
<fencedevices>
<fencedevice agent="fence_bladecenter" ipaddr="XXXXX" login="USERID" name="BladeCenterFencing" passwd="******"/>
</fencedevices>
...
Of another cluster is same except node names differ(they blade3-1 and blade3-2).
2.
In /etc/hosts
192.168.154.1/24 blade4-1.privdomain blade4-1 (this goes on eth1 of blade4-1)
192.168.154.2/24 blade4-2.privdomain blade4-2 (this goes on eth1 of blade4-2)
Same is with blade3-1 and blade3-2 (Cluster 2)
192.168.153.1/24 blade3-1.privdomain blade3-1 (this goes on eth1 of blade3-1)
192.168.153.2/24 blade3-2.privdomain blade3-2 (this goes on eth1 of blade3-2)
3. Public IP of blade4-1 & blade4-2 (Cluster 1) and blade3-1 & blade3-2 (Cluster 2) are on same subnet 10.5.81.xxx and subnet mask is 255.255.255.
Thanks
On Wed, Mar 28, 2012 at 12:54 PM, Digimer <lists@xxxxxxxxxx> wrote:
I'll admit to having trouble following you... So let's get some basics;
* What is the cluster.conf of either cluster?
* What exact IPs does the hostnames, as set in the cluster.conf, resolve to?
* What is the exact IP of eth0 and eth1 on each node, inc. netmask?
As a point of curiosity; why do you need the cluster names to be identical?
Digi
On 03/28/2012 12:00 AM, Parvez Shaikh wrote:
> Thanks Fabio & Digimer,
>
> Just to add more information that each node has 4 NIC cards out of which -
>
> eth0 is public / management IP their addresses are falling on the same
> subnet
>
> eth1 is private address and they are privately connected (node A - node
> B) and (node C - node D)
>
> Cluster.conf mentions hostname(from /etc/hosts) which corresponds to IP
> addresses of eth1 (not globally accessible name of eth0's IP)
>
> Even in situation such as above, cluster names should be unique? I would
> assume that ccsd would use only the node names given in cluster.conf for
> communication.
>
> Does ccsd use cluster name to locate other members of cluster and get
> copy of cluster configuration file from those?
>
> Following messages are seen in /var/log/messages -
>
> Mar 28 11:57:07 blade4-2 ccsd[6821]: Starting ccsd 2.0.115:
> Mar 28 11:57:07 blade4-2 ccsd[6821]: Built: Mar 16 2010 10:28:57
> Mar 28 11:57:07 blade4-2 ccsd[6821]: Copyright (C) Red Hat, Inc. 2004
> All rights reserved.
> Mar 28 11:57:07 blade4-2 ccsd[6821]: cluster.conf (cluster name =
> Cluster, version = 1) found.
> Mar 28 11:57:07 blade4-2 ccsd[6821]: Remote copy of cluster.conf is from
> quorate node.
> Mar 28 11:57:07 blade4-2 ccsd[6821]: Local version # : 1
> Mar 28 11:57:07 blade4-2 ccsd[6821]: Remote version #: 1
> Mar 28 11:57:07 blade4-2 ccsd[6821]: Remote copy of cluster.conf is from
> quorate node.
> Mar 28 11:57:07 blade4-2 ccsd[6821]: Local version # : 1
> Mar 28 11:57:07 blade4-2 ccsd[6821]: Remote version #: 1
> Mar 28 11:57:07 blade4-2 ccsd[6821]: Remote copy of cluster.conf is from
> quorate node.
> Mar 28 11:57:07 blade4-2 ccsd[6821]: Local version # : 1
> Mar 28 11:57:07 blade4-2 ccsd[6821]: Remote version #: 1
>
>
> On Wed, Mar 28, 2012 at 12:13 PM, Fabio M. Di Nitto <fdinitto@xxxxxxxxxx
> <mailto:fdinitto@xxxxxxxxxx>> wrote:> Linux-cluster@xxxxxxxxxx <mailto:Linux-cluster@xxxxxxxxxx>
>
> On 3/28/2012 8:14 AM, Parvez Shaikh wrote:
> > Hi experts,
> >
> > I am running in to a problem in a situation where two clusters in the
> > network have same name.
> >
> > Node A, Node B : cluster name CLUSTER
> > Node C, Node D : cluster name CLUSTER
> >
> > Node C and Node D's cluster is running fine however when I start
> node A
> > it copies /etc/cluster.conf (I believe ccsd) from either C and D and
> > replaces its own and start attempting a cluster, of course it fails
> > because C or D is not a host name of node A.
> >
> > Is it possible to have more than one clusters in same network, these
> > nodes are in IBM blades hosted in Bladecenter and the heart beat
> IPs are
> > on private network.
>
> Uh.. you can have multiple clusters on the same network as long as they
> have different names. Or you need to isolate them on different networks
> if you need to keep the same name.
>
> Fabio
>
> --
> Linux-cluster mailing list
> https://www.redhat.com/mailman/listinfo/linux-cluster
>
>
>
>
> --
> Linux-cluster mailing list
> Linux-cluster@xxxxxxxxxx
> https://www.redhat.com/mailman/listinfo/linux-cluster
-- Linux-cluster mailing list Linux-cluster@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/linux-cluster