dmesg errors

[Ken Scharpell]

Ive got the following errors from dmesg. Whats the problem and whats the recommended solution ?

SCSI: Reset detected -- lbolt: 89101, bus: 1
SCSI: Reset detected -- lbolt: 90696, bus: 1
lbp->state: 4020
lbp->offset: ffffffff
lbp->uPhysScript: 200000
From most recent interrupt:
ISTAT: 02, SIST0: 02, SIST1: 00, DSTAT: 80, DSPS: 0020001c
lsp: 0000000000000000
lbp->owner: 0000000000000000
scratch_lsp: 0000000000000000
Pre-DSP script dump [00000000410000e0]:
e0340004 00000000 e0100004 00000000
48000000 00000000 78350000 00000000
Script dump [0000000041000100]:
50000000 0020001c 80000000 0000000b
0f000001 002005c0 60000040 00000000

SCSI: Read error -- dev: b 31 0x01e000, errno: 126, resid: 2048,
blkno: 8, sectno: 16, offset: 8192, bcount: 2048.
LVM: vg[0]: pvnum=0 (dev_t=0x1f01e000) is POWERFAILED
LVM: Recovered Path (device 0x1f01e000) to PV 0 in VG 0.
LVM: Restored PV 0 to VG 0.

[Stefan Farrelly · Accepted solution]

Ive seen this a lot of times and in my opinion your drive powered down then up again. Once it powered down the first thing HP-UX detected was SCSI errors trying to talk to the disk (Resets) then it kept retrying and now produced a SCSI read error, and then after more retrying if finally correctly deduced the disk had powered itself off, but it did eventually power itself back up again and everything returned to normal.

Ive seen other opinions that this is caused by the IO timeout length being too small (pvchange -t) but IMHO this is incorrect - its always caused by a physical disk problem. In the unlikely event the default IO timeout is too small then the SCSI errors in dmesg look different to what you have above.

IMHO your disk is dying. This will continue to happen until it dies completely or you replace it. I would call HP and get it replaced.