Ignite failed and cannot restart

[ralph barber]

I tried to clone a system from an ignite server.
The server had previously been ignited with an image of a different server. The make recovery failed the message in the logs was
* Testing pax for needed patch xx
x x * Passed pax tests. xx
x x * Checking Versions of Recovery Tools xx
x x ERROR: Cannot stat device file: /dev/vx/dsk/dg01/vol3: No such file or xx
x x directory (errno = 2). Check /etc/fstab for a bad entry. x
x x ERROR: The list_expander command failed. This could be due to a proble x
x x the file: "/var/opt/ignite/recovery/client_mnt/0x00306E4B1F32/re x
make_net_recovery completed unsuccessfully

However in the ignite GUI it says it is 5% done and "Active with Warnings". I cannot see any ignite processes running on the target or ignite server The trouble is I can't restart or run a new install or cancel the one I tried to run so I'm stuck
HELP !

[MarkSyder]

The clue is in errno 2: this indicates a non-existent file. In this case, device file /dev/vx/dsk/dg01/vol3.

You say you've ignited from an image of a different server. It appears that the other server had a different number of discs and ignite is trying to read from or write to a non-existent disc.

I don't think you'll be able to ignite from this image. Have you an alternative one from a server with the same number of discs?

Mark Syder (like the drink but spelt different)

[Eric Antunes]

Hi Ralph,

Check if you have this /dev/vx/dsk/dg01/vol3 volume stated in you /etc/fstab file, remove it and try again!

Best Regards,

Eric

[ralph barber]

I thought of commenting out the fstab entries but the problem still remains that I cannot start a new install as it seems to think the one that failed is active. Is there a flag I can reset to resolve this ?

[MarkSyder]

It should time out.

On the server which runs ignite, check the file /var/opt/ignite/server/ignite.defs - the value client_timeout is the one you are looking for. According to the manual page, it is typically set to 30 minutes.

Mark

[ralph barber]

Mark The value is set to 20 minutes
I ran the ignite 18 hours ago so it doesn't look like the timeout is going to kick in for some reason ???

[MarkSyder]

The server you are attempting to boot is not doing anything productive at present, so it will not do any harm to reboot it.

The ignite server should detect that the remote server is not responding and stop the process. If that doesn't work, I think we'll have to consider more drastic action (kill -9?) but we really should save that for the last resort.

Mark

[Eric Antunes]

Try to restart the Ignite server...

[ralph barber]

I tried rebooting the target server this morning. I cannot reboot the IGNITE server, it is not in my control and probably is far too busy to contemplate lightly.

[MarkSyder]

Where are Pete Randall, SEP, and Bill Hassell when you need them? Given the time difference between UK and USA, probably still in bed - or in a voting booth.

If you have a support agreement I think it's time to make a phone call to HP. If not, I would be looking for processes on the ignite server that you can kill:

ps -ef|grep ignite

If I have time to look in my HP-UX book over the coming minutes I'll try to come up with something more productive than killing processes.

Mark

[Michael Roberts_3]

On the Ignite server look in
/var/opt/ignite/clients//recovery/latest/recovery.log
to see what the last message written to the server was. Also on the server you can run
'nfsstat -m' and check for problems with NFS mounts.

Closing down the GUI will remove the read() on the log file, and that may release the processes.