topic Re: dmesg in Operating System - HP-UX

dmesg

Ravinder Singh Gill — Fri, 26 Aug 2005 08:42:04 GMT

when I did dmesg I got the following as part of my output. Can anybody advise what this is & what action should be taken?

SCSI: Read error -- dev: b 31 0x003000, errno: 126, resid: 2048,
blkno: 8, sectno: 16, offset: 8192, bcount: 2048.
LVM: vg[1]: pvnum=3 (dev_t=0x1f003000) is POWERFAILED

SCSI: Write error -- dev: b 31 0x003000, errno: 126, resid: 43008,
blkno: 1238, sectno: 2476, offset: 1267712, bcount: 43008.
percival:/#

Re: dmesg

Patrick Wallek — Fri, 26 Aug 2005 08:48:33 GMT

Check your disks and see if you have a disk that has died.

The worriesome message is:
LVM: vg[1]: pvnum=3 (dev_t=0x1f003000) is POWERFAILED

That indicates the c0t3d0 has potentially failed.

Re: dmesg

Torsten. — Fri, 26 Aug 2005 08:49:12 GMT

Hi Ravinder,

that seems to be a bad disk.
Try

ioscan -fnkCdisk

and look for "NO_HW" to find it.

Re: dmesg

Jeff Schussele — Fri, 26 Aug 2005 08:49:30 GMT

Hi Ravinder,

It means that at some point - dmesg is not time stamped, check the syslog.log for date/time - the system saw a problem with /dev/dsk/c0t3d0
This might have been a transient problem, but at the time the system could not access it.

HTH,
Jeff

Re: dmesg

Alex Lavrov. — Fri, 26 Aug 2005 08:51:46 GMT

Yout have a bad disk in your system.

Look at this line:

dev: b 31 0x003000

If you do "lsdev | grep 31", you'll see that 31 is number of disk driver, so the failed device is disk.

To find the disk that failed, you must look at "0x003000". I guess the disk is /dev/dsk/c3t0d0.

Do, "ioscan -fnCdisk" and look at the lines that have "NO_HW" in them. If you see it, than defentely the disk is gone. To verify it, do:

dd if=/dev/rdsk/c3t0d0 of=/dev/null bs=8192

You probably will see error messages or dd will hang. Call HP support and replace the disk.

Alex.

Re: dmesg

Alex Lavrov. — Fri, 26 Aug 2005 08:53:06 GMT

Sorry, the disk is dev/rdsk/c0t3d0 ofcourse.

My mistake.

Re: dmesg

Cem Tugrul — Fri, 26 Aug 2005 08:55:45 GMT

Ravinder,

What about the lights on this disk?
amber/green?

it seems something wrong on your power supply
of the disk.

ioscan -fnCdisk?
(As Torsten mentioned) check state of disk
CLAIMED or NO_HW

Good Luck,

Re: dmesg

John E.Ophious — Fri, 26 Aug 2005 08:58:23 GMT

G'day mate,

Just as all my other partners have posted, it looks like you could potentially have a bad disk. However, I would like to add that before tossing it in the trash, you may want to examine the disk for any loose connections/bent pins/missing terminators, etc. I've had this happen in the past where a simple reseat would solve the problem. If this does work, get a good backup asap and be sure to monitor this disk closely afterwards for any performance degradation.

Re: dmesg

Ravinder Singh Gill — Fri, 26 Aug 2005 08:59:10 GMT

i did an ioscan -fnC disk and all the disks showed as claimed.

Re: dmesg

Siju Jose_1 — Fri, 26 Aug 2005 09:08:40 GMT

Hi Ravinder

I would suggest you run a cstm .Look for errors in the output and REPLACE the drive at the earliest.

a.cstm
b.map --select the device Num
c.sel dev "dev no"
d.info
e.infolog

Siju

Re: dmesg

Pat Lieberg — Fri, 26 Aug 2005 09:08:59 GMT

You could do a strings on /etc/lvmtab and see if that device is in use. Perhaps for some reason the device files exist for a disk that is no longer in the system?

If its an in-use disk, I woud try doing the dd test Alex mentions above.

If it seems to be working now, I would watch if closely. This could be a sign of impending failure.

Re: dmesg

Bill Hassell — Fri, 26 Aug 2005 09:09:13 GMT

ioscan does not test disks, it just requests a SCSI identification. Yoiu definitely have a bad disk and not only dmesg but syslog are probably full of errors. Make sure you have a good backup. If you have disk mirroring, replacement is fairly easy.

Re: dmesg

Mel Burslan — Fri, 26 Aug 2005 09:18:16 GMT

It is beneficial for you to check the syslog for the same type of powerfail messages. Sometimes they are immediately followed by power restored to PV such and such messages. Even if the power to the PV is being restored, it is a good indication that there is somethign wrong with the hardware. A service call is definitely in order for you.

Re: dmesg

Ravinder Singh Gill — Fri, 26 Aug 2005 09:19:21 GMT

I checked ioscan again and this time the disk is showing as NO HW. I have placed a call with HP as they provide us support. Thanks guys. I will take your advice on board and keep looking into it myself. If you have any other tips please let me know. Thanks

Re: dmesg

Alex Lavrov. — Fri, 26 Aug 2005 09:21:42 GMT

Hmm :) Assigning some points to the ppl that helped you the most will be nice ;-)

Alex.

Re: dmesg

Ravinder Singh Gill — Fri, 26 Aug 2005 09:22:08 GMT

What does the dd command that Alex mentions above try to do exactly?

Re: dmesg

Alex Lavrov. — Fri, 26 Aug 2005 09:26:14 GMT

dd reads byte by byte from the "if" and writes it to "of". That's input file and output file. This was you check that you can read the whole disk.

If the disk is faulty, you'll probably get errors from dd, that it couldn't read some block.

By the way, that's the first thing HP ask you to do.

Alex.