Disk Errors in log files

[Bona_1]

Hi all,I am getting some weird errors in my logfiles
The server is an RP5470 connected to a DS2300 disk system. Can anyone make head or tail out of the errors?

***********************
Jan 26 17:40:32 billing vmunix: LVM: vg[12]: pvnum=0 (dev_t=0x1f101500) is POWERFAILED
Jan 26 17:40:33 billing vmunix:
Jan 26 17:40:33 billing vmunix: SCSI: Async write error -- dev: b 31 0x101300, errno: 126, resid: 8192,
Jan 26 17:40:33 billing vmunix: blkno: 50104696, sectno: 100209392, offset: -232398848, bcount: 8192.
Jan 26 17:40:32 billing vmunix: LVM: vg[12]: pvnum=0 (dev_t=0x1f101500) is POWERFAILED
Jan 26 17:40:33 billing vmunix: blkno: 49292056, sectno: 98584112, offset: -1064542208, bcount: 8192.
Jan 26 17:40:33 billing vmunix: blkno: 49204920, sectno: 98409840, offset: -1153769472, bcount: 8192.
Jan 26 17:40:33 billing vmunix: blkno: 49204928, sectno: 98409856, offset: -1153761280, bcount: 8192.
Jan 26 17:40:33 billing vmunix: SCSI: Write error -- dev: b 31 0x101300, errno: 126, resid: 16384,
Jan 26 17:40:33 billing vmunix: SCSI: Async write error -- dev: b 31 0x101300, errno: 126, resid: 8192,
Jan 26 17:40:33 billing above message repeats 3 times

***********************************

Thanks

[RAC_1]

Seems that the disks /dev/dsk/c1t5do and /dev/dsk/c1t3d0 have problems.

Can you pvdisplay on these disks and check.
How these disks are connected? Do they have alternate paths?

[Karthik S S]

Try installing the latest SCSI patches for HP-UX (the version you are running). You can download it from,

http://www1.itrc.hp.com/service/patch/pdbMainPage.do

-Karthik S S

[Ravi_8]

Hi

Looks like disks have gone for a toss,

check the disks using diskinfo and dd for any errors.

#diskinfo /dev/rdsk/cxtxdx
#dd if=/dev/rdsk/cxtsdx of=/dev/null bs=1024K

[Fabio Ettore]

Hi,

I agree with Kartik!
I faced this problem few days ago.

I think that your problem is solved by installing the following patches:

PHKL_29527 - s700_800 11.11 Filesystem buffercache performance fix
PHKL_28513 - s700_800 11.11 SCSI IO Cumulative Patch

If it doesn't solve the problem then you should follow hw checks on disks /dev/dsk/c1t5d0 and /dev/dsk/c1t3d0.

Good luck!
Ettore

[Kent Ostby]

FYI, the devt listed above ( dev_t=0x1f101500 ) can be correlated to a specific disk by doing :

ll /dev/dsk | grep 101500

[RolandH]

Could it be that you have a heavy I/O load on the DS2300?

Then it could be that you must increase the IO_Timeout from your Physical Volume. That can you do with the pvchange command. The default is 30 sec. But I recommend you not increase it higher than 60 sec.

# pvchange -t 60
That can be done online

Doc

[Steven E. Protter]

If timeout does not fix this, eventually you will end up needing to replace that disk.

write errors are not tolerable for production servers.

SEP