should I set the TIMEOUT on these 2 disk?

[Kathy McGohan_1]

We have several HP-NClass 4000s with 12H AutoRaids. Every now and then, I see messages like the ones below. Just need some information why I keep getting these messages, please. Thanks.

Jun 1 02:18:58 dmc20 vmunix: SCSI: Request Timeout; Abort Tag -- lbolt: 109333413, dev: 1f030000, io_id: 321b21c - (This is disk c3t0d0)
Jun 1 02:18:58 dmc20 vmunix: SCSI: Request Timeout; Abort Tag -- lbolt: 109333413, dev: 1f041100, io_id: 464a9fc - (This is disk c4t1d1)
Jun 1 02:18:59 dmc20 vmunix: scb->cdb: 2a 08 01 40 81 80 00 00 10 00
Jun 1 02:18:59 dmc20 vmunix: scb->cdb: 2a 00 04 e3 61 b0 00 00 10 00
Jun 1 02:18:59 dmc20 vmunix: scb->cdb: 2a 00 04 e3 33 80SCSI: Request
Timeout; Abort Tag -- lbolt: 109333513, dev: 1f030000, io_id: 321b21b
Jun 1 02:18:59 dmc20 vmunix: 00 00 40 00
Jun 1 02:18:59 dmc20 vmunix: scb->cdb: 2a 00 04 e3 66 80 00 00 40 00
Jun 1 02:18:59 dmc20 vmunix: scb->cdb: 2a 00 04 e3 4a 30 00 00 50 00
Jun 1 02:18:59 dmc20 vmunix: scb->cdb: 2a 00 04 e3 4a 80 00 00 10 00
Jun 1 02:18:59 dmc20 vmunix: scb->cdb: 2a 00 04 e3 5a 00 00 00 30 00
Jun 1 02:18:59 dmc20 vmunix: SCSI: Request Timeout; Abort Tag -- lbolt: 1093335
13, dev: 1f041100, io_id: 464aa02
Jun 1 02:19:00 dmc20 vmunix: SCSI: Abort Tag -- lbolt: 109333513, dev: 1f041100, io_id: 464a9f8
Jun 1 02:19:00 dmc20 vmunix: SCSI: Abort Tag -- lbolt: 109333513, dev: 1f041100, io_id: 464a9fb
Jun 1 02:19:00 dmc20 vmunix: SCSI: Abort Tag -- lbolt: 109333513, dev: 1f041100, io_id: 464a9fb
Jun 1 02:19:00 dmc20 vmunix: SCSI: Abort Tag -- lbolt: 109333513, dev: 1f041100, io_id: 464aa05
Jun 1 02:18:59 dmc20 vmunix: scb->cdb: 2a 00 04 e3 61 b0 00 00 10 00
Jun 1 02:19:00 dmc20 above message repeats 5 times
Jun 1 02:19:00 dmc20 vmunix: SCSI: Abort Tag -- lbolt: 109333513, dev: 1f041100, io_id: 464a9fe
Jun 1 02:19:00 dmc20 vmunix: SCSI: Abort Tag -- lbolt: 109333513, dev: 1f041100, io_id: 464a9ff
Jun 1 02:19:00 dmc20 vmunix: SCSI: Abort Tag -- lbolt: 109333513, dev: 1f041100, io_id: 464aa09
Jun 1 02:19:00 dmc20 vmunix: LVM: PV 0 has been returned to vg[1].
Jun 1 02:19:08 dmc20 vmunix: LVM: PV 0 has been returned to vg[2].

[Stefan Farrelly · Accepted solution]

For starters, your Autoraid disks are very slow - unfortunately one of the slowest disk subsystems ever made by HP, so its likely they are just being overloaded with requests.

Try increasing the timeout value using pvchange -t 60 (or more) and see if that fixes it.

Other possibilities are that a disk in the Autoraid is powering down then off- check the Autoraid for errors.

Try checking the queue depth on the disk devices (see man on scsictl) using the command; scsictl -m queue_depth /dev/rdsk/cXXXX
You can try increasing this also, allowing more devices to queue up I/O requests.

[A. Clay Stephenson]

Hi Kathy,

I typically set my AutoRAID's to a timeout of 120 seconds via pvchange.

Another thing that helps is to set up the LUN's so that about half of the I/O is thru controller X (alternate Y) and vice versa. If you simply use SAM to setup an AutoRAID and don't use the 'Use Hardware Path' options then all the I/O will go thru only 1 of the 12'H controller unless X fails.

Clay