lvextend is stuck

[Dan Newcombe]

(lets try this again!!!)

bash# lvextend -l 2550 /dev/vg01/play /dev/dsk/c0t12d0 /dev/dsk/c1t12d0
Logical volume "/dev/vg01/play" has been successfully extended.

That is what I entered and the response I got. It has not come back to a prompt yet. There doesn't appear to be any abnormal disk activity going on, and everything else is running fine.

I can't kill the process or anything...it looks like a reboot is the only way to clear this up.

Any ideas as to how to clear this up without a reboot or what could have caused it?

[Rammig Claus]

Hi Dan,

it seems that your lvextend is hanging in kernel mode (perhaps hanging on IO).
Therefor the only possibility to kill this process is a reboot.


Best regards ...
Claus

[John Poff]

Hi,

Do you get any results from 'lvdisplay /dev/vg01/play' ?

JP

[Dan Newcombe]

Yup...it got hung too :)

I can't kill those off either.

Lot of cursing going on over here.

[Jeff Schussele]

Hi Dan,

I assume you pvcreated those disks & vgextended the VG successfully....

What was the LE count prior to adding these disks?

Are these TWO seperate disks or alternate paths to the same device?

Is this LV mirrored?

Disks known to be good?

What happens if you hit enter a couple of times? Ctrl-C?

I'd pull up another root session & check to see if the process is actually running & what it's Parent PID is & try killing that (if NOT 1) before I'd reboot.

Rgds,
Jeff

[John Poff]

Any clues in syslog?

I'm curious. Are c0t12d0 and c1t12d0 really two different PVs, or is one a PV link?

JP

[Sridhar Bhaskarla]

Hi Dan,

For me, it looks like you have a bad disk.
(could be one of them that you used in the command line) Your dmesg /var/adm/syslog/syslog.log should be giving out errors. Do they?

Yes. The situation is cursable. But I suggest you find out the reason first before rebooting the box.

-Sri

[Dan Newcombe]

Okay...lots of questions.

Before adding, the LE count was 2253, tring to increase it to 2550 to match our production database LE count.

These are two seperate disks on two controllers and should be mirrors of each other. (yes...the LV is mirrored). I'm beginning to wonder if I goofed in expecting it to extend to the c0t12d0 and mirror to the c1t12d0 - maybe this will get people to stop making requests right before I leave :)

As far as I know the disks are good. I don't recall off the top of my head what else is on them, but have not seen any problems.

Pressing enter does what you think it would - moves things up one line. Ctrl-C or Ctrl-Z does nothing.

I haven't killed it's parent id yet, namely because when I did that with the hung lvdisplay and pvdisplay, it just killed the parent process and reparented the other processes to 1...maybe I should kill 1 :)

No clues in syslog, but I had to restart that (/var filled up recently which freaked syslog out)

These are two seperate disks.

dmesg does have a SCSI read error, but it is for another disk (based on the device number given) Looks like the whole machine is falling apart.

Right now I'm doing a backup to tape, just in case after I reboot it comes back pretending it doesn't know about any LV's.

(BTW - just killed the parent process of lvextend which reparented it to init.

Just wish these K boxes didn't take 20 minutes to POST.

[Dan Newcombe]

Looks like the drive c0t9d0 was down for some reason. I reseated the drive and it is fine now. The drive was one of the ones that was in the logical volume in question.

Guess it should have told me that something wasn't kosher instead of locking up when trying to do the extend.

I'm gonna keep a very close eye on that drive.

[Manju Kampli]

good to see u have figured out .. the lvextend comamnd was hanging on IO .. this you can see thro glance .. you could have seen some errors in syslog.log file OR dmesg command...

its always good to look around for any clues

[Dan Newcombe]

I think part of the problem was that syslog had taken a fall earlier so that is why it missed the io errors.

I will be keeping a very close eye on this drive and syslog.

It seems that our machine that has been acting flawlessly, sans a memory carrier problem, is starting to show it's age :(

Thanks for everyone's ideas and help...is there a way of recovering from lvextend hanging on I/O like this? Perhaps reseating the drive?