nospace - /dev/root - 1 extent left

[OldSchool]

I've started getting the above message (or very similar), but only occassionally, and I've been trying (unsuccessfully) to track down the cause.

/dev/root is NOT mounted, and all of the filesystems look good, both when running bdf or du -k.

HP-UX 11.0 on K-class, LVM, no online JFS,

I did get the message once when a large ftp transfer was taking place, and once during the "shutdown" phase of a reboot.

Any ideas?

Thx...Scott

[Pete Randall]

If a file had filled up /, been removed, but was still open by the process writing to it, / would continue to grow - that's the only scenario I know of that could cause you to continue to get the message.


Pete

[Jaime Bolanos Rojas.]

Sys Adm,

Even thought you might have already checked it this thread might give you some guidance:

http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=1003681&admit=-682735245+1155045772144+28353475

Regards,

Jaime.

[Mel Burslan]

mentinoning of /dev/root makes me think that when this system was initially brought up, it came up in LVM maintenance mode and then comands like initi 3 or init 4 might have been run to take it up to normal operation mode.

As Pete said, there might have been some data streams left attached to /dev/root device at that time which were continuing to grow.

If you can afford about 30 minutes of down time, I'd suggest rebooting the system.

[OldSchool]

Mel,

The system WAS in LVM maintenance mode several months ago, and has been rebooted twice since to install patches.

What puzzles me, is why the message(s) refer to "/dev/root" when it is not, and has not, been mounted since the lvm maintenance was complete.

[OldSchool]

Lets try this again, as I got an authentication error when submitting:

Pete:

Yep, I knew that one. BTW: its /dev/root that is in the message. I know this usually comes from lvm maintenance if you don't bounce the box

Mel:

The server WAS in LVM maint / single user mode, but that was about a year ago. Box has been rebooted at least twice since then, plus one extended period (12+ hours) of power-off time, due to electrical problems with the building and generators.

What's got me confused is its talking about /dev/root. I know it's not mounted, as I would have had to mount it myself.

[Pete Randall]

Scott,

I think that's just a function of the way the error message refers to the file systems. It doesn't say that /var is full, for example. It says /dev/vg00/lvol9. In the case of root, I suspect it just refers to at as /dev/root. If you had come directly out of LVM maint mode, bdf would show it as /dev/root as well.


Pete

[OldSchool]

No /dev/root in bdf. checked fstab/checklist, and no dice.. it's not mounted

I can to a "bdf /dev/root" and get the results as if I had done "bdf /" (except for the name of course).

So, maybe root filled up. One user was ftp'ing a large file into directory under /, and during the shutdown, I suppose the log file write could have tripped the message.

[Pete Randall]

Well, if you're not still getting the message, bdf reports OK, du looks good, and the system is still running, I would say you're seeing a historical event in the log and it's not still full.


Pete

[OldSchool]

The problem reared its ugly head Friday, during the ftp session, and once again Monday morning when we did a shutdown -h so I could pull a couple of cards out.

It's hasn't occurred since the system came back up.

Its more of a head scratcher than anything.

Thx..