Stale LV

[sara1]

After a mirrored root disk replacement it shows the following error:

LV Name /dev/vg00/lvol7
LV Status available/stale
LV Size (Mbytes) 436
Current LE 109
Allocated PE 218
Used PV 2


Can anybody suggest the issue.

[Devender Khatana]

Hi,

This is because you have not yet done vgsync or lvsync after replacing disk.

#lvsync /dev/vg00/lvol7

Will fix it. Also if other LVOLs are also stale then you can also do

#vgsync /dev/vg00

HTH,
Devender

[Devender Khatana]

Hi,

Apart from this you can get a list of stale LE's using

#lvdisplay -v /dev/vg00/lvol7 |pg

Also to get the status of other LVOLs use

#lvdisplay -v /dev/vg00/lvol* |pg

After syncing the LVOL using lvsync all the LEs should get current and you should not see stale.

Also you can check this by grepping stale like
#lvdisplay -v /dev/vg00/lvol* |grep -i stale |pg

HTH,
Devender

[Arunvijai_4]

Hello,

You can use lvsync /dev/vg00/lvol7
and # lvdisplay -v /dev/vg00/lvol7

-Arun

[Arunvijai_4]

This could be because of bad blocks in your disk. Just try this out,

# dd if=/dev/vg00/rlvol7 of=/dev/null

-Arun

[sara1]

Devender
Thanks for the reply.
But
lvsync /dev/vg00/lvol7
lvsync: Couldn't re-synchronize stale partitions of the logical volume:
I/O error
lvsync: Couldn't resynchronize logical volume "/dev/vg00/lvol7".

[Sivakumar TS]

Hi Sara,

please do
#lvdisplay -v

and check whcih PV is showing/having stale Extents.

Is that the disk you have replaced ?

You can do a "dd" test to confirm the disk (media) error.

do

#dd if=/dev/rdsk/ of=/dev/null

With Regards,

Siva.

[Arunvijai_4]

Hi, check these threads for more solutions,

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=640938
http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=153123

-Arun

[Devender Khatana]

Hi,

This means that the device is faulty. This could be even the new disk you put in or the other disk which was left from the initial mirror.

The same device can be found by
#lvdisplay -v /dev/vg00/lvol7

The device having LE's marked as stale is the one having problem. Also find out how many of LE's are shown as stale in the output.

dd solution will not work here atleast at LVOL level as the read will be performed from the other copy and you will not get any error in the output. A better idea will be to do the same read test on the suspected disk device and find out the results. If the no. of stale LEs is quite high, the disk should be replaced immediately.

#dd if=/dev/rdsk/cxtydz of=/dev/null bs=2048k

Where /dev/rdsk/cxtydz is the suspected disk.

HTH,
Devender

[sara1]

Number of stale extents is only two. So is there any other way out other than doing a disk replacement.

[sara1]

dd if=/dev/rdsk/c0t5d0 of=/dev/null bs=1024k
dd read error: I/O error
4080+0 records in
4080+0 records out

[Devender Khatana]

Hi,

How many of the LEs are free on the disk.

#vgdisplay /dev/vg00

If there are some free PEs on the disk, you can do a workaround by mirroring it on remaining PEs. But this is only workaround as the disk showing I/O error on some part may extend the same to some more area anytime and can not be assumed to be reliable.

Also what is the sequence of the LE's that are marked stale.

Attach complete output of
#lvdisplay -v /dev/vg00/lvol7

&

#vgdisplay -v /dev/vg00

HTH,
Devender

[sara1]

HI,
Please find attached.

[Devender Khatana]

Hi,

All the LEs on the disk are allready allocated in such case you will have to reduce some LVOL atleast by 4 LEs to accomodate this LE's. is that feasible for you and which LVOL you can reduce. Also post the output of bdf to find with which LVOL we can have the option of reducing.

Also if reducing is to be done without additional "Online JFS" software it requires rebooting to singe user for some LVOLs. My best advice will be to go ahead with replacing the disk as it was recently it should be still under warranty.

Also do assign points to the earlier responses to your threads.

HTH,
Devender

[Ajitkumar Rane]

Hello Sara,

Since only 2 PE are marked stale, probably your disk has bad blocks, rather it has started developing bad blocks. Better to get the disk replaced as may be the bad block count will keep increasing and finally the mirror will fail.

Other way round is break the mirror, remove the disk from the VG, format the disk and reconfigure, but this is not a good option for an unreliable disk.

Thanks

Ajit

[sara1]

We dont have a valid mirror now so that we can replace that disk. So whats the way out now.

[Devender Khatana]

Hi,

You do not have the valid mirror now but you still have good original copy. The new mirror will be created onto the new disk and this is the normal way of replacing the disk in case of a mirror when one copy get lost or corrupted due to hardware problem.

Is this what was your question?

HTH,
Devender

[Ajitkumar Rane]

Sara,

You should unmirror the lvols if they are mirrored at all.
lvreduce -m 0
vgreduce
replace the disk
go as per the procedure for adding a root mirror.

Hope this helps

Rgds,

Ajit

[Devender Khatana]

Hi,

Is it resolved? If yes could you post the actual procedure adopted.

HTH,
Devender

[Manikantan]

The dd command has clearly given you an output stating that the disk has gone bad (I/O error).

It is better to change the disk because

1/ It is lvol7 which is mostly /usr in HPUX and a critical FS too.No one would like to take a risk with this FS.

[Devender Khatana]

Hi,

Is it resolved? Atleast update the status and the method followed.

Also if this was helpful, do remember to assign points to the responses to the thread. It will help others in future to find solution for similar prolems and will get you good responses later on.

HTH,
Devender