scsi read error /lbolt scsi error

[yusuf iqbal]

Hello friends,

we have a r390 which after a reboot came up with the following error:

SCSI: Resetting SCSI -- lbolt: 13117, bus: 0
SCSI: Reset detected -- lbolt: 13117, bus: 0
vmunix: LVM: Recovered Path (device 0x1f006000) to PV 0
in VG 0

the device is actually the root disk cot6d0, I used the utility cstm to verify the output in syslog. It came back successfull. Please advise/help guys!

thanks in advance

[Stefan Farrelly · Accepted solution]

Its not too fatal, you just received a warning that the connection to your disk was temporarily lost. Either due to a faulty cable, controller or disk. Most likely the disk. LVM recovered aok.

If these errors continue (check dmesg or syslog frequently) then your disk may be on its way out and you may need a replacement. Contact HP who will want to see these SCSI errors before deciding if a replacement is needed.

[Rita C Workman]

This can be because of a couple things.

But considering that it's your root disk...I would hope your mirror/disk (and that you have root mirrored..) is bootable and I would get that disk replaced ! Cause once it starts..I tend to not wait for the other shoe to drop..

Just my thoughts,
Rita

[BFA6]

I would agree with Rita, you should get this swapped out.

It could be that the disk is on the way out.
Keep checking syslog.log to see if these are a regular occurrence.

Hilary

[Helen French]

Just my thoughts:

This error can occur due to number of things

1) a firmware error, scsi controller, cables or disk hardware error. I would check and confirm these with an HP engineer.

2) Patches - Apply the latest patches to the system. Checke for latest SCSI, LVM and vxfs patches.

3) If you want to change the SCSI time-out value of the disk, then try this command:

# pvchange -t 180 /dev/dsk/cxtydz ( i think you can go up to -t 300 max.)

[T G Manikandan]

If that is your root disk it requires a immediate mirror for that .
Also you can create a bootable tape using ignite which can help you in case of trouble.

Also check your terminator and the controller cables.
Do that before you get a I/O hang.

Also check this link

http://forums.itrc.hp.com/cm/QuestionAnswer/0,,0x354368c57f64d4118fee0090279cd0f9,00.html

Thanks

[steven Burgess_2]

Hi Yusuf

Try a dd to see if you get any I/O errors

dd if=/dev/dsk/c0t6d0 of=/dev/null bs=1024k

You can set a timeout for scsi resets

Have a look at man pvchange

-t option I think

Also

echo 2408,10?2x|adb /dev/dsk/c0t6d0

Should also return clean

See you in a couple of weeks

Steve

[yusuf iqbal]

friends,

I created an ignite tape staright after i recived those messages yesterday and I also changed the default time out of the disks from 0 to 60, using the pvchange command. I checked the syslog this morning and there were no new messages, however as most of my replys mentioned earlier I will need to verify the initial error messages in the case of a disk failure. Its better to be safe than sorry!!!

thanks guys!

[yusuf iqbal]

Guys,

Good job I did the the ignite yesterday, as the root disk crashed! :( I have arranged for hardware to be replaced and will look at recovering the server.

Thanks to all who gave feedback to my Questions!

thanks