How can I fix stale extents?

[pap]

Hi,
There are some stale extents in my /stand, /root logical volumes. Can anyboy tell me how can I fix this?

Thanks,
-pap

[James R. Ferguson]

Hi:

# vgsync /dev/vg00

Regards!

...JRF...

[Helen French]

Hi Piyush:

# vgsync

this will do the task. Also find out if you have some errors or why this was happened ?

HTH,
Shiju

[Darrell Allen · Accepted solution]

Assuming you are mirroring with MirrorDisk/UX:

vgsync vg00

-or-

lvsync /dev/vg00/lvol1
lvsync /dev/vg00/lvol3

Darrell

[Sridhar Bhaskarla]

Hi,

"vgsync vg00" does the job. But if the mirror disk is bad , then it will give out errors. You will need to find if there is any problem with the disk.

-Sri

[Clemens van Everdingen]

Hi,

If you have stale extents and this is the first time you experience them, then ou have to wonder how old the disks are.
If they are running for a long tine already, I would have them changed for new ones.

Else if they are mirrored, try the vgsync suggestion the other already gave.

Regards,
C.

[Tom Danzig]

If the vgsync doesn't work, do an lvsync on each lvol with stale extents.

[S.K. Chan]

I would be "slightly" concern if I got stale extents though it doesn't necessarily represent a 100% hardware problem. Find out which PV is having the stale PE and you might want to run dd check on that PV or check system logs (syslog, dmesg, STM) to make sure no hardware problem is detected.

[Jon Finley]

You may also be looking at file system damage. If the above mentioned vgsync/lvsync's don't resolve the issue, you may need to run fsck on stand and root.

Another possibility is that the disk is starting to go (hopefully you're running in a mirrored configuration). One test you can perform is to issue a dd against the questionable disk.

If it's c0t0d0:

dd if=/dev/dsk/c0t0d0 of=/dev/null

If you know about where the staqle extents are, you can add in a count parameter to limit how much of the disk dd will read.

Regards,

Jon