Metrocluster => quorum server will move, what should I change

[wim merckx]

Hello guys,

first timer here, I became responsible for a metrocluster, but unfortunatly I didn't had any training about this, so actually I don't know anything about it :-). Anyway, tomorrow they're going to physically move the quorum server (diff IP address) so what files should I modify?:
1) on the metrocluster itself
2) on the packages
3) ...

Help me out guys!
Cheers
Wim.

[Berlene Herren]

This may help...

http://www.docs.hp.com/hpux/onlinedocs/B7660-90001/B7660-90001.html

Berlene

[wim merckx]

thx, but I couldn't find specific details on which files to change when the quorum server receives another IP address...

I really need to know which files to modify...

Cheers
Wim

[Michael Steele_2]

First, the quorum server CAN'T BE ON A DIFFERNT SUBNET than the nodes. So is it just the 'quorum server' moving or all the nodes in the cluster? If just the quorum server then you've got problems. See this link:

http://docs.hp.com/cgi-bin/fsearch/framedisplay?top=/hpux/onlinedocs/B5158-90044/B5158-90044_top.html&con=/hpux/onlinedocs/B5158-90044/00/00/28-con.html&toc=/hpux/onlinedocs/B5158-90044/00/00/28-toc.html&searchterms=server%7cquorum&queryid=20030422-092404

Second, you also have to take the heartbeat ip into consideration since this is automatically sensed by the 'cmquerycl' command at cluster creation time. The first 'qualifying' LAN as well as all subsequent alternates LANS can all be designated to carry the heartbeat during failover. Because of this I believe you're looking at a deletion of the current cluster file and a recreation of a new cluster file. This is because new ip's are being introduced and the 'cmquerycl -C 'qualifications' will have to be rerun. Network latency for example is considered.

Are your hostnames also changing? Then the package.conf file will also have to be updated and this requires the 'cmmakepkg', 'cmcheckconf -P' and 'cmapplyconf -P' commands.

This is a long procedure and I'm not sure if I can properly convey all the nuances without misinterpretation. I KNOW I WOULD GO OVER THIS WITH PAPER AND PEN a couple of times before starting. But as you are pressed for time I would get on the horn to the response center and try to enable their assistance. But in general:

make_tape_recovery = two
cp /etc/cmcluster/* /tmp/.
cmhaltcl
cmdeletecl
update all host ip changes via 'set_parms initial' command on all nodes in the cluster.
update .rhosts or cmnodelist.
cmquerycl -c file -n node1...
cmcheckconf -C file
cmapplyconf -C file

update package.conf files
cmmakepkg -P package.conf
cmcheckconf -P pa...
cmapplyconf -P pa...

This is probably the most difficult procedure there is in SG, by the way. Is it too late to hire an SG consultant?