Looks like a disk failed.

[Steven E. Protter]

SCSI: Reset detected -- lbolt: 25984, bus: 1
SCSI: Unexpected Disconnect -- lbolt: 21448284, dev: 1f018000, io_id: 1131aec
LVM: Recovered Path (device 0x1f058500) to PV 3 in VG 1.
LVM: Restored PV 3 to VG 1.


Just wondering if someone can give me a trick on how to translate this to a /dev/dsk/c number.

I'm looking for a formula so I can do it myself next time.

The disk is mirrored no biggie.

SEP

[Khalid A. Al-Tayaran]

Hi,

Do an ioscan -fnC disk

If one disk is dead you;'' get a NO_HW under the s/W state meaning no hardware copy the device file name (character) then use:

diskinfo -v /dev/rdsk/cXtYdZ

Then if you don't get any info the disk is either stuck or dameged...

[Steven E. Protter]

Turns out the disk isn't dead. A reboot cleared the message.

I'd still like a simple formula on how to translate the message into a /dev/dsk device.

SEP

[john korterman · Accepted solution]

Hi,
may this document helps?
http://www4.itrc.hp.com/service/cki/docDisplay.do?docLocale=en_US&docId=200000063234348

(Hope you get the same result as for other suggestions)

regards,
John K.

[Michael Steele_2]

dev: 1f018000

Indicates c1t8d0.

Always the middle string of two followed by zeros.

Not always effective, fibre for example.

[S.K. Chan]

Breifly those are in hex.
1f = major number (decime 31 ie "sdisk" - from lsdev output)
01 = card instance # (bus number)
8 = scsi id
0 = device LUN
00 = (this I forgot what it represent).
What I usually do is just look in my /dev/dsk and you should see a matching device for 018000. That way you don;t need to bother translating it.
# ll /dev/dsk|grep 018000

[Steven E. Protter]

Great suggestions. Thanks.

I don't know why I haven't picked that up in class over the years.

Close the thread. Thanks.

SEP

[Eugeny Brychkov]

Steve,
sometimes when disk is queried to read some its surface areas read operation can fail (you know it of course). But disk do not tell host immediqately "I'm bad" but simpy trying recovering data for too long thus causing timeouts and link switching.
As I guess this disk you received disconnect for is fibre channel one, isn't it? Although it's fibre channel, it works using SCSI commands and requests are passing SCSI driver. You know that SCSI Id 8 has the lowest priority, so all devices located on the 'SCSI bus' will have a priority on this disk (target)
Eugeny

[Steven E. Protter]

Hello Eugene. It was actually a regular old scsi disk this time. So far the message has not returned.

Since everything on those local disks is Raid 1/0, I'm going to monitor and deal with it after my Passover vacation is over.

SEP

[Pradeep_3]

Can you check PV time out usind,

pvdisplay -v /dev/dsk/cXtYdZ | pg

Normally pv time out value should be 180 recommended by HP. This may also cause the problem.

You can change pvtime out value online using

pvchange -t 160 /dev/dsk/cXtYdZ

Observe some time and check for error getting repeated at same interval.

If still you are facing same problem,
you can try
dd if=/dev/dsk/cXtYdZ of=/dev/null bs=1024

This will check your disk integrity. If successfully completed then media of disk is o.k.

If disk is o.k. ceck the cable, termination statistics and finally check for patched for SCSI IO or LVM patches.