[Linux-cluster] Two-node setup Help (RHES5)
nlam87346 at library.usyd.edu.au
Thu Sep 20 23:58:01 UTC 2007
On Thu, 2007-09-20 at 15:45 -0400, Ozarchuk, John D wrote:
> Hi all,
> I am trying to configure a two-node cluster that will simple be NFS,
> FTP, and SMB fileshares. I cannot see to get each node to join the
> cluster. I am fencing via the ilo2 ports, and that portion is
> working. However, when I try to bring up the services on both nodes
> they say “fence failed <hostname>”.
> I am not sure why the nodes are trying to fence each other, probably
> because neither of them can join.
> I can resolve hostnames back and forth to each node, they can ping one
> another, and they are both on the same VLAN.
Maybe it's overkill, but it's pretty straight-forward to check whether
each of your nodes is receiving multicast cluster packets from the other
node using tcpdump. Assuming you've got everything on eth0 and your
other node's IP address is 192.168.1.100 you would use (with root privs)
tcpdump -v -nn -i eth0 ip multicast and src 192.168.1.100
You should see some packets from your other node sent to the default
cluster multicast group IP address, which is probably 18.104.22.168.
More information about the Linux-cluster