Replacing primary boot device & a corrupted alternate device query

[Joanne Keegan]

Hi Everyone,

I'm trying to understand/work something out that happened not long ago.

A server booted up on the alternate boot device. The primary was replaced (hot swappable). Stand was mirrored back on to the primary device, and a part of root. The mirroring process was killed, and the system rebooted. This was done on the understanding the the mirror devices would update each other based on the timestamp on files on either device. This did not work but corrupted the alternate boot device.

The server was then recovered via a make_recovery tape.

What I'm interested in is the process of rebooting from a primary root volume that has partial data from VG00 and expecting the alternate to update it. Has anyone done this? Is it supposed to work, amd if so, how?

Hopefully I've explained it clearly enough.

Any help will be appreciated.

Regards,

Jo

[harry d brown jr]

Jo,

>>> process of rebooting from a primary root volume that has partial data from VG00 and expecting the alternate to update it. <<<

It doesn't work this way.

>>> Has anyone done this? Is it supposed to work, amd if so, how? <<<

A mirror is exactly that: "a mirror"

When you have a "primary", and I mean PRIMARY as in the MAIN device running the system. So if your OLD_PRIMARY root disk fails and you are running on your OLD_ALTERNATE (which is NOW your NEW_PRIMARY), then you replace the disk of OLD_PRIMARY, and MIRROR the NEW_PRIMARY (the OLD_ALTERNATE) back to the new disk of OLD_PRIMARY.

God I hope I didn't confuse you! That's a lot of double speak!

live free or die
harry

[Joanne Keegan]

Hi Harry,

I certainly do not disagree with you. I am trying to understand what an HP engineer did onsite, and the logic he used. While it didn't work with what he was trying to achieve, I wonder if he was mistaken, or this has highlighted a bug or something.

Regards,

Jo

[harry d brown jr · Accepted solution]

Funny, I have never meet a "field engineer" that I would actually allow to swap my mirrors, of course that's my way of doing things.

Jo, I think they were mistaken or confused. Of course I got one of my first big jobs back in 1982 because someone thought they knew how move data from one disk to another.

Next time, you DRIVE, they navigate, and YOU question EVERYTHING and have them get clarification from their level 3 support people.

live free or die
harry

[Patrick Wallek]

If the system was up and running, and the drives were resyncing, I am really confused as to why the system was rebooted.

The basic steps to take when replacing a hot swap drive: pvcreate commands, cgcfgrestore, and vgsync. Once the sync is completed, you can reboot to test the drive, but if it is a production system, there are other ways to check that the mirror was successful.

I have had HP field service folks cause me grief when replacing a mirrored disk, because they weren't 100% sure what they were doing and their "cookbook" had the recipe all wrong. I wound up having to restore about 40GB of data that I shouldn't have had too.

I now know better!!!!!! And I now replace all disks myself!!

I would never recommend sycing part of a VG and then rebooting to sync the rest. Especially with VG00. Once the sync starts, let it finish.

[Joanne Keegan]

I agree with all of the replies. Thank-you for your input.