Scsi error

[mukesh_a]

Hi All i have following dmesg in hp-ux , can somebody let me know problem with which disk and what is the way to find out the faulty disk .

LVM: vg[0]: pvnum=0 (dev_t=0x1c013000) is POWERFAILED
LVM: vg[1]: pvnum=0 (dev_t=0x1c010000) is POWERFAILED
LVM: PV 0 has been returned to vg[0].
LVM: PV 0 has been returned to vg[1].
dev 1c010000 got invalid b_resid 0 b_bcount 40960, error = 0x7e
dev 1c010000 got invalid b_resid 0 b_bcount 24576, error = 0x7e
dev 1c010000 got invalid b_resid 0 b_bcount 32768, error = 0x7e
dev 1c010000 got invalid b_resid 0 b_bcount 32768, error = 0x7e
LVM: vg[1]: pvnum=0 (dev_t=0x1c010000) is POWERFAILED
LVM: PV 0 has been returned to vg[1].

SCSI: Request Timeout -- lbolt: -411203889, dev: cd003000
lbp->state: 4020
lbp->offset: ffffffff
lbp->uPhysScript: 280000
From most recent interrupt:
ISTAT: 22, SIST0: 04, SIST1: 00, DSTAT: 80, DSPS: 00000006
lsp: 1c012080
bp->b_dev: cd003000
scb->io_id: 8135e0
scb->cdb: 10 00 00 00 00 00
lbolt_at_timeout: -411263989, lbolt_at_start: -411263989
lsp->state: 10d
lbp->owner: 1c012080
scratch_lsp: 0
Pre-DSP script dump [cd8e030]:
78346000 0000000a 78351000 00000000
0e000002 00280500 80000000 00000000
Script dump [cd8e050]:
9f0b0000 00000006 0a000000 00280508
721a0000 00000000 98080000 00000001

SCSI: Abort -- lbolt: -411203889, dev: cd003000, io_id: 8135e0

[J. Bravo]

Hello:

To find the broken disks, search for the minor numbers 013000 and 010000 under /dev/dsk or /dev/rdsk and then you'll see the device file as following:

ll /dev/rdsk/|head
total 0
crw-r----- 1 bin sys 188 0x006000 Jun 18 14:44 c0t6d0
crw-r----- 1 bin sys 188 0x0a0000 Sep 21 2006 c10t0d0

With this information you can identify the broken disks with ioscan -fnC disk | grep cXtXdX and in the output of ioscan you'll find the hardware path of the HBA. For expample, if ioscan shows 0/0/2/0.2.0, then your HBA is 0/0/2/0.

Regards;

J. Bravo.

[Aneesh Mohan]

Hi ,

Scsi errors within the syslog are usualy a good indication that there is a hardware issue with a diskdrive.


Do

#ll /dev/dsk |grep minor number




/Aneesh

[Steven E. Protter]

Shalom,

multiple alerts as you have received sometimes indicate a controller or drive cage failure. Hardware is most likely going to at least need some attention.

SEP