Crash cascade with ignite

[David Cruañes Lee_1]

We had a crash cascade with:

HP-UX ADLDBP01 B.11.00 U 9000/800 101901577 unlimited-user license
(+ cluster manager with oracle+broadvision)

We still don't now the cause of the first crash but the following appeared to be a problem with ignite...we made a mistake in the middle of the crash:

We were using the following command to make the tape:
/opt/ignite/bin/make_recovery -iv >> $LOG

And a hp-ux engineer told us that we should use:

/opt/ignite/bin/make_recovery -C -A -iv >> $LOG

But we used the same tape after the first recover so now the ignite tape seems to be useless. That will explain the crash cascade.

Is there any other useful document for crash recover to follow the patch&soft depot install and cluster manager?
Which is the correct make_recovery command?

Greetings!


#man make_recovery
(...)
-A
(...)
In the case of large root disks or volume groups, or when other utilities are used for normal backups, this option should not be used. Instead, the default minimum core OS should be backed up, to recover a minimum system, and then the full recovery should be done from the backups.

[Steven E. Protter · Accepted solution]

Suggestion:

Stop using obsolete make_recovery

use make_tape_recovery

That might work better.

SEP

[Mark Greene_1]

Your make_recovery syntax is incomplete. Try this:

make_recovery -A -v -p -d /dev/rmt/xxx

make_recovery -r -v -C -d /dev/rmt/xxx


The first step is a preview and builds files that you can edit to customize your ignite, including custom recovery scripts. I use this to auto-mirror my drives after reboot. The second step is what actually creates the tape.

mark

[David Cruañes Lee_1]

Thanx Steven & Marc !!

Anyway, what can we do with this tape?
Is any info useful in it to recover the system structure (cm conf, net conf, disk array or whatever was in the server)?

It is possible to read the first ignite backup?
Best regards!

[Mark Greene_1]

From the make_recovery man page (on my 11.0 system):

To extract a single file from the recovery archive:

mt -t /dev/rmt/0mn fsf 1
tar -xvf /dev/rmt/0m filename

Extracting files from tape can take a long time, especially when
archives are large.


mark

[David Cruañes Lee_1]

thanks Marc!

But that seems a long, long way to bring up the server. I think I will keep forward with the normal restore...see attached file.

Anything else important to keep in mind??

(And I hope this weekend I will relax myself with some caipirinhas and nice apt-get-install commands; mixing up solars,hp-ux and microssft at the same job is unhealthy...just keeding... ;)