[Linux-cluster] usedev directive not working correctly?

[Kovacs, Corey J.]

Sorry about the previous message subject, too lazy to type the address and
didn't change the subject.

On one hand I agree with that, however I've gone as far as to set up static
routes for the addresses and lock_gulmd won't start at all since it can't
talk to the other lock servers at all. As I said in the original message,
'gulm_tool nodelist node1' reports that the lock manager on node3 is NOT
using the directed interface but node2 and node1 are.

But, in the interest of fixing this I'll separate the second nics onto their
own vlan and try again. 

I've copied the original message here so there will be less confusion.


=======================ORIGINAL MESSAGE BEGIN=============================
I have been trying to get GFS (6.0.2.20) to use the second nic by using the
"usedev" directive to no avail. Actually, what happens is the third node
always uses the first nic. My config looks like this...
 
/etc/hosts
==========================
192.168.1.11      node1
192.168.1.12      node2
192.168.1.13      node3
192.168.1.101    node1-ic
192.168.1.102    node2-ic
192.168.1.103    node3-ic
 
cluster.ccs
==========================
cluster {
                name = "mycluster"
                lock_gulm {
                        servers = [ "node1" , "node2" , "node3" ]
                }
}
 
nodes.ccs
==========================
nodes    {
    node1 {
        ip_interfaces {
                    node1-ic = "192.168.1.101"
        }
        usedev = "node1-ic"
        fence {
                iLO {
                        node1-ilo {
                                    action = "reboot"
                        }
                }
        }
 
    }
    node2 {
        ... (omitted for space)
    }
 
    node3{
        ... (omitted for space)
    }
 
} 
 
 
When I start lock_gulmd on all the nodes, the first two (node1, node2) report
(via gulm_tool nodelist node1) they are using the second nic as directed, but
node3 always reports nic1 being used. The first nics are all connected to
100M links and the seconds nics are all 1GB links. I can ping all the nic2's
using "ping -I eth1 node1-ic" etc and they all respond. I've even gone so far
as to set up static-routes but that just killed things because node3's
lock_gulmd keeps binding to nic1.
 
 
Any ideas as to what I might be doing wrong?
=======================ORIGINAL MESSAGE END=============================




-----Original Message-----
From: linux-cluster-bounces at redhat.com
[mailto:linux-cluster-bounces at redhat.com] On Behalf Of Birger Wathne
Sent: Tuesday, June 07, 2005 10:35 AM
To: linux clustering
Subject: Re: [Linux-cluster] Can I use cman instead of theHA
(HighAvailability) package?


> /etc/hosts
> ==========================
> 192.168.1.11      node1
> 192.168.1.12      node2
> 192.168.1.13      node3
> 192.168.1.101    node1-ic
> 192.168.1.102    node2-ic
> 192.168.1.103    node3-ic

I think it would be far better to use a separate subnet for the Gb links. It
would make it much easier to know that the packets get routed correctly.

--
birger

--
Linux-cluster mailing list
Linux-cluster at redhat.com
http://www.redhat.com/mailman/listinfo/linux-cluster