harddisk error - plex detach - volume vol3, plex vol3-01

[Narayanan_2]

Recently I have created a LSM volume using 3 mirrored disk. I found this error in /var/adm/messages,
Jul 7 19:48:38 mysmpalpha vmunix: io/vol.c(volerror): Uncorrectable write error on volume vol3, plex vol3-01, block 16829296
Jul 7 19:48:39 mysmpalpha vmunix: voliod_error: plex detach - volume vol3, plex vol3-01
The harddisk size has reduced totaling only 2 disk. What do i have to do to fix this?

[Narayanan_2]

Sorry, gave the wrong info... The harddisk size did not reduce, it remained the same size.

[Ralf Puchner]

if you see a write error, why no checking binary.errlog for possible hardware issues?

[Nagarajan Balakrishnan_1]

Narayanan,

You have mentioned that one of the mirrors have failed in the three mirror volume. As mentioned by Ralf, check the logs, and identify the disk.

If the disk has completely failed, then it would be straight forward to replace and using voldisk command to add this into the LSM and re-mirror (second mirror).

Else, you need to look into the errors coming out of the disk and decide whether it warrants replacement or not. You can check with your support engineer.

Cheers
Baalki

[Dmitry Timoshenko]

Hi

I think, that it`s better to try to restore disk:
#volplex -g att vol3 vol3-01

This will take some time, but there is no strick relations between hardware failure and LSM plex errors.
If it fails, you can ask HP support for help.

Best regards,
Dmitry.

[Ralf Puchner]

it very dangerous to restore disk in case of a hardware problem, maybe some data will be lost so it is important to isolate the hardware problem, changing disk and after that restoring disk.

If it was a single BBR you can goon with mirroring.

[Dmitry Timoshenko]

Dear Ralph,

Why this operation is dangerous?
We`re just trying to read data from healthy plex, right?

Please, explain this with more details.

Thanx in advance,
Dmitry.

[Ralf Puchner]

it is dangerous if you try to mirror all plexes without checking if an hardware error exists. And binary.errlog is the key to do the job.

[Dmitry Timoshenko]

Ralph,

Sorry, but this is not clear enough.
Why this operation is dangerous?

Dmitry.

[Ralf Puchner]

because if the hardware problem is not analyzed the mirroring could lead to an intact volume but with damaged disks or content leading to additinal problems or to a problem if the current active plex will fail or must be synced.

So why not analyzing the binary.errlog first prior to recreate the mirror?

[Dmitry Timoshenko]

May be this is stupid, but i have no idea how to analyze this file without support of HP. I have to wait for a long time for response from our local HP office.
It is extremely dangerous to keep our data without mirroring for a long time.

Dmitry.

[Ralf Puchner]

if you have no knowledge of SCSI return or status code you must wait for an analyze, but the HP support center will responde in time.

The other method is to check the binary.errlog output and find your own answer by learning scsi codes and registers. But using a damaged disk as part of a mirror is not very clever. You can also swap the disk to a new one without analyzing - maybe the fastes way if not using a spare set which indicates your data are not really save...

[Narayanan_2]

Thanks all for your input. I have resolved the problem by contacting HP engineer and replacing the faulty hd. Just for info, before i contacted HP engineer, i did uerf | -100 and found some scsi errors. After that i tried to do volpex mirror the harddisk again but it failed. Finally did reboot and the hd automatically started mirroring. After complete mirroring there were the same messages again.