mirror and re-mirror of vg00

[Johnny Damtoft]

Hi all,

I've just setup 2 identical servers for MC/SG Cluster.

Itanium HP-UX version 11.23.

First i installed nodeA, did all configuration etc to complete the server installation, including vg00 mirror.

After the mirroring, i split the vg00 mirror, and put the "mirror" disk in the server nodeB.

Note: I've been using the HP Document "When Good Disks Go Bad".
ftp://ftp.hpux.dk/unix/hpux/docs/When_Good_Disks_Go_Bad.pdf

nodeB was started up, I changed the hostname, IP addresses etc.

On both nodeA and nodeB i re-mirrored vg00, to make everything perfect for the cluster configuration.

Then i had 2 identical servers, with different IP and Hostname - and all mirrored up.

Then i restarted nodeA, and disaster happened.
- I could not boot from one of the disks, and when i was trying to boot from the alternate disk i was toled that the quorum disks is missing.

When starting up in Single User Mode, i can see that everything looks find, so i booted in LVM Maintenance mode. (hpux /stand/vmunix -lm).

After a little work, i could communicate with my vg00 again (still in Single User Mode), and when i execute vgdisplay -v vg00, everything looks perfect.

Now i have one BIG wish.

I want to verify, repair and if possible, re-mirror ALL my vg00 disks so I can be 1000% sure that everything is working perfectly.

Please show me the way, and commands to use.

Please advice.

thanks :)


-- Johnny Damtoft

[Ivan Krastev]

See this document - http://docs.hp.com/en/5992-1910/5992-1910.pdf

for complete procedure.

regards,
ivan

[Johnny Damtoft]

-- story continued --

Please find the attached error message in TXT file, from nodeA


-- Johnny Damtoft

[Torsten. · Accepted solution]

Well, it's a sas disk.

The quorum disk is missing, so you need to boot in quorum mode first.

Stop the boot process:

(C) Copyright 1999-2006 Hewlett-Packard Development Company, L.P.
All rights reserved

HP-UX Boot Loader for IPF -- Revision 2.028

Press Any Key to interrupt Autoboot
\EFI\HPUX\AUTO ==> boot vmunix
Seconds left till autoboot - 9

press any key here!

HPUX> boot vmunix -lq

[Torsten.]

How did you "split" and "remirror" on nodeA?

[Johnny Damtoft]

I have also attached the output from: strings /etc/lvmtab, ioscan -fnCdisk, vgdisplay -v vg00 and lvlnboot -v vg00.

** ** ** **

Yes, it's 2x SAS disks.

Once i've booted in -lq mode, what's next?

[Johnny Damtoft]

Torsten:
I did not do the split my self, someone else did. :(

- And they simply went to the server, and unplugged the disk without pvchange -a n first :(

This task gave me a little noice when booing on nodeB, but I managed.

[Torsten.]

Are the commands performed on nodeA or nodeB?

[Geoff Wild]

First, this is NOT the way to build a cluster.

To make them identical, you should have used ignite to build an image of the first node, then used that image to create the second node - that way you get 2 identical systems and you also test your disater recovery method.

Seeing how you are on Itanium, the steps to mirror root are a bit different:

http://www.met.ca/itrc/index.php?option=com_content&task=view&id=34&Itemid=2

Rgds...Geoff

[Johnny Damtoft]

Torsten:
All of this is beeing performed on nodeA.

The server nodeB is currently running the cluster package, and can not be touched.

[Torsten.]

You say your server crash while booting - when did you run these commands?

Once it is up in quorum mode, run the commands again.

The system was booting from 0/2/1/0.0.0.1.0 - so there was a problem with the other disk 0/2/1/0.0.0.2.0 or (more likely) with the config.