topic Re: available/stale? in Operating System - HP-UX

available/stale?

Hanry Zhou — Mon, 07 Apr 2003 15:49:44 GMT

Hi,

I got following message, when I run vgdisplay on:
LV Name /dev/vg02/lvol2
LV Status available/stale
same message on lvol3.

However, I don't have any problem to access lvol2. What causes stale problem, and how to remove the message?

Thanks,

Re: available/stale?

Patrick Wallek — Mon, 07 Apr 2003 15:55:03 GMT

Is /dev/vg02/lvol2 mirrored? If it is then you have a disk that is probably going bad or has already gone bad.

Do a:

# vgdisplay -v /dev/vg02

At the very bottom where it lists the disk, see if one of the disks is unavailable. If it is then you will need to get the disk replaced, make it part of the VG again and resync your mirrors. Once the mirrors have been re-synced, you will see the status as "available/syncd".

Re: available/stale?

John Poff — Mon, 07 Apr 2003 15:57:09 GMT

Hi,

Do an 'lvdisplay -v /dev/vg02/lvol2' and you'll see all the LEs for that logical volume. You should be able to see which ones are stale. If you aren't having any disk problems, you can just do a 'vgsync vg02' and that should resync things. I'd check into any possible problems, though. Having stale extents can be a sign of trouble. Maybe you are having some SCSI issues with a PV, such as the drive is starting to go bad, or maybe a cabling issue?

JP

Re: available/stale?

Ramkumar Devanathan — Mon, 07 Apr 2003 15:58:10 GMT

see man vgdisplay.

LV Status is stale when "Logical volume is available but contains physical extents that are not current".

- ramd.

Re: available/stale?

James R. Ferguson — Mon, 07 Apr 2003 16:02:25 GMT

Hi:

This indicates that you have a mirrord logical volume which is operating *as intended*. That is, you are able to read and write data even though you have lost (temporarily or "permanently" the ability to access all or part of one of the mirrors).

It is possible that one of the physical disk to which the logical volume's extents map, has failed or is failing.

Try resynchronizing the logical volume and/or the whole volume group:

# lvsync /dev/vg02/lvol2

# vgsync /dev/vg02

Since you see stale extents on lvol3, too, I'd try resynchronization at the volume group level.

Regards!

...JRF...

Re: available/stale?

Michael Steele_2 — Mon, 07 Apr 2003 16:11:35 GMT

Better note the device and HW address for this disk.

vgdisplay -v | more

ioscan -fnkC disk (* for HW address *)

-or-

lssf /dev/rdsk/cXtYdZ

Then test the disk.

dd if=/dev/dsk/cXtYdZ of=/dev/null count=1000000
control c (* to break *)

Note: You need the count=1000000 to get past the cache.

Also check STM for I/O errors. You'll need the HW address now.

STM > TOOLS > UTILITY > RUN > LOGTOOL > FILE > VIEW > RAW SUMMARY.

Note the first and last dates of transactions and calculate the difference. If the difference is short, like 4 hours, then this is important to note. Now read down the report of hardware addresses and observe the integer numbers in parenthesis. Anything over 150 in this 4 hour period should be called into HP for replacement.

Finally, a lot of people don't mirror their swap and this is a mistake for there is no "...swapoff..." command. Swap is reduced by commenting out its entry in /etc/fstab and rebooting. In this case if you lose a disk that has swap on it the system will panic and reboot itself.

Consequently, get your backups in order especially your make_tape_recovery tapes. Take a couple.

Re: available/stale?

Hanry Zhou — Mon, 07 Apr 2003 16:48:51 GMT

How do I locate the bad disk name, /dev/dsk/c?t?d?, all I got are "???" on the PV1/PV2 columns from lvdisplay

Re: available/stale?

Patrick Wallek — Mon, 07 Apr 2003 17:00:58 GMT

Do a 'vgdisplay -v /dev/vg02' and post the results, please.

Re: available/stale?

Michael Steele_2 — Mon, 07 Apr 2003 18:44:36 GMT

It???s a trial and error test of disks in vg02. First obtain a list from vgdisplay -v. Then,

dd if=/dev/dsk/cXtYdZ of=/dev/null count=1000000
control c (* after a count of 6 or so. *)

If the prompt doesn't return it???s a bad disk.

Also use the STM > LOGTOOL procedure.

Just read the above posting carefully and execute.

Re: available/stale?

Michael Steele_2 — Mon, 07 Apr 2003 18:48:04 GMT

Sorry, vgdisplay won't work with ???.

Try

strings /etc/lvmtab

and note the disks under the vg02 entry.

-or-

vgcfgrestore -f /dev/lvmconfg/vg02.conf -l

This will also work

Re: available/stale?

Hanry Zhou — Tue, 08 Apr 2003 12:57:38 GMT

Thank you all.

We found one of disk in Jamica enclosure had problem to acces, and we replaced it. However, strangely I can not find the disk drive in xstm window. The HW is 8/4.3.0(c1t3d0), I can find it in the result of ioscan -fnC disk, but not in xstm window?