make_recovery tape

[fg_1]

All

I just ran into a situation yesterday trying to run a make_recovery -A -C -v.

Our root drive crashed on an RP-7400, the 0/S is mirrored, all lvol's were available/stale.

At my supv's request I tried to run a make_recovery tape with the above command. To my discovery the job hung up and did not run, I gave it 2hrs to complete, but the process was not getting any cpu time.

I was wondering if it's at all possible to run a make_recovery and have it write the data from the mirror drive instead of the primary root disk by altering some configuration file that points to the drive.

Any replies are appreciated.

Thanks.

[Ian Dennison_1]

I may have the wrong idea here, but if you have suffered a failure of the root disk, should you not be restoring from a make_recovery, rather than generating one?

Or did you not have an initial make_recovery of the system?

Cheers, Ian Dennison

[Krishna Prasad]

I have seen this problem with Ignite when one of root disk died. After we replaced the bad disk and did the standard LVM vg00 recovery of mirrod disk the Ignite worked.

I have upgraded my version of Ignite since so I don't know if it was a problem with the old version or the basic behavior of Ignite?

[fg_1]

Let me just clarify, we were creating a make_recovery tape just as a precaution, we were not doing a restore of vg00.

I am just trying to figure out if your primary drive is out can you configure ignite_ux to run a make_recovery using a mirror copy of the root drive.

[harry d brown jr]

Frank,

The only ways to backup your mirrors are:

dd if=/dev/dsk/c#t#d# of=/dev/rmt/sometapedrive

or

dd if=/dev/vg##/lvol## of=/dev/rmt/sometapedrive

or

mount them, by first splitting the mirrors, then backuping up the filesystems. But if they are STILL MIRRORED, then you don't need to do this!

Did the first make_recovery finish, meaning you don't already have one running?

Is your system totally up and running now?

live free or die
harry

[fg_1]

harry

the make_recovery was started around 9:30AM, and by the time I got back about 11:40AM it was still running which is unusual, usually only takes 1hr. I had to abort it. The system is operational with all lvol's in vg00 in an available/stale state.

I assume by your statement that the dd is the only way to do this, not with make_recovery?.

[harry d brown jr]

Frank,

You are running from your mirrored disk (lv's), correct? Have you replaced your failed disk?


live free or die
harry

[fg_1]

harry

not yet, the chicken's that run the application area dont want us touching the box.

I am running off the mirror at this time only. No Redundency.

[Sanjay_6 · Accepted solution]

Hi Frank,

Make recovery will not work if the primary disk has failed. This is because it tries to read the lif info from the primary disk and is hung up over there.

you cannot use make_recovery to create a backup tape at this point of time. Replace the primary disk and then rebuild the mirror and then run make_recovery.

Hope this helps.

Regds

[harry d brown jr]

Frank,

Do a tar backup (or whatever you use) of your system. You already have a good make_recovery tape, correct? With the two backups you can safely "recover" your system.

as for the make_recovery issue, that's strange. The only thing I can think of is that maybe your mirror isn't bootable (no LIF)?


live free or die
harry

[Helen French]

Hi Frank,

I would suggest you to first replace your bad disk and re-mirror it ! You can do the make_recovery after that. It is always better to make the mirror copy functional, instead of making a dd. Once everything is back to normal, for safety you can do either dd or make_recovery.

Also while u r doing make_recovery now, it will give you some problems, because the primary disk is unavailable. The VG information on your mirror disk has indications about your primary disk !

HTH,
Shiju

[Krishna Prasad]

Frank,

Well I got my earlier question answered too. Ignite will not do a backup when the primary drives fail even though the mirrors are active.

My next question is are your vg00 drives hot plugable? If they are why won't the application people let you fix your own box?
I understand if it is an internal drive that requires the box to come down. Most newer systems however are hot plugable drives.

[Corthouts Carlo]

Hi,

You could try running the make_recovery command first in preview mode.

make_recovery -p -d

This will create a file in the /var/opt/ignite/recovery directory.
The file is named config.recover.

Here you can change somethings which might help you with making a recovery tape from your mirrordisks.

first you have a line like :
==
init _hp_root_disk="HW_path"
==

Also check the _hp_disk_layout.
It also contains some variables about the physical disk.

You could try and experiment to check if you
are able to make a recovery tape.

To create the tape when all has been changed just use the make_recovery -r command.

[Darrell Allen]

Hi Frank,

I don't know for sure but I suspect Sanjay has nailed it. Make_recovery is trying to get info from the disk you booted up from which is now toast. The mirrors are keeping you running and you haven't rebooted, right?

You could probably reboot off the alternate boot disk and create a make_recovery. However, I'd get the disk replaced and my mirrors re-synced ASAP. You're okay unless something happens to the alternate disk also.

Harry's idea to make a tar (or fbackup, etc) is good.

Darrell

[fg_1]

All

The consensus i get is that if the root disk is down, make_recovery is about as useless as a pair of >>>>> on a nun. (ha ha).

Bottom line was that we are going to replace the bad drive and then make it bootable, then re-sync vg00 and boot off the primary disk again.

I have backed up the vg00 mirror volumes using omniback 2 just to cover my six.

Thank you for all your answers, points are being assigned.

This is the best group on the net.

[Sanjay_6]

Hi Frank,

The reason i said, make_recovery is trying to copy the lif info from the primary root disk is that when the system is hung trying to create a make_recovery archive, do a "ps -ef" and it tells you the make_recovery is trying to get the lif info from the primary disk. I has a similar situation earlier when i tried to create a make_recovery archive before i was replacing the primary root disk that had failed.

Hope this helps.

Regds