911 Emergency System using Service Guard

[daniel figueroa]

We are in the process of trying to recover our primary node from a corrupted disk.

This configuration is not a good one and will be replaced in a matter of weeks. It has no mirroring of the root drives and is running HP-UX 10.20. But until then, I have to deal with this problem:

After running a make_recovery_tape to restore the primary from a corrupted root drive, we brought the primary online. The cluster support engineer changed the network hostname for what ever reason. The alternate node is running fine for the dispatch folks to handle their 911 calls, but as soon as the cluster specialist brings the primary online, the alternate is downed on the network. There goes our dispatch folks from handling 911.

So, right now we have disconnected the NICs and can review the configuration to troubleshoot this problem. We had to reboot the alternated and reset the 911 software. For lanadmin would not allow us to reset the card.

Your help is appreciated.
Daniel

[Jeen Oppedijk]

Daniel,

It looks like there are 2 identical network adresses at the network when you start the primary. I would start comparing both network configurations....
Jeen

[Stephen Doud]

If the server was restored from a make_recovery tape created for another system (something that HP doesn't officially support), then it is appropriate to change not only the systems' hostname, but every parameter that makes this server unique - such as the link level addresses if modified(see /etc/rc.config.d/hp*conf) and LAN IPs (/etc/rc.config.d/netconf).

-s.

[daniel figueroa]

corrections were made in the /etc/cluster configuration files by editing in the hostname change of the primary. The cluster was reformed and is running fine now.

Thanks for your help.
Daniel