HPE 9000 and HPE e3000 Servers
1748194 Members
4273 Online
108759 Solutions
New Discussion юеВ

K-Class SCSI Critical Error

 
SOLVED
Go to solution
Tim Krego_1
Frequent Advisor

K-Class SCSI Critical Error

First let me say I am fairly new to HP/UX and their 9000 servers. I have inherited the daily duties for our K460 machine. Our sysadmin/dba quit.

I took HP/UX for Experienced Admins last month, but I am still learning.

My problem:

In the syslog yesterday I have a vmunix: SCSI request timeout. Critical error on 10_0.6.0. I think that is one of the internal 4GB Seagate OS disks.

There is an AutoRAID with all the slots filled with 9GB disks. The error is not on any of those disks.

Right now I do not know if the 2-4GB internal disks are mirrored. I do not think we have any type of HP support, but I need to find out how to purchase support from HP.

The syslog.log said to run a resdata command to find out the error. I have included that file in this message if anyone can look at it and point me in the right direction.

Thanks,
Tim Krego
HP/UX Newbie
8 REPLIES 8
A. Clay Stephenson
Acclaimed Contributor
Solution

Re: K-Class SCSI Critical Error

Tim,
I can't be sure but 10.0.6 is probably your boot disk so it is vital to get this fixed soon. Because it is a boot disk you will need to make sure you have a good backup before disk replacement.
1) ioscan -C disk -fn | lp
This will give you a list of attached disks
2) vgdisplay -v /dev/vg00
Look for the Physical Volumes sections:
If you have only 1 volume listed then you are not mirrored. If you have more than 1 then you MAY be mirrored. At this point also look
at the disk devices associated with vg00 and compare them to your ioscan listing. You should then be able to find 10.0.6 and know that it is in vg00. (If not then do a vgdisplay of vgxx until you find it. I think it's vg00).
3) If you found multiple physical volumes for vg00 then note the locigal volumes listed with the vgdisplay.
4) For each logical volume do an lvdisplay
e.g. lvdisplay /dev/vg00/lvol3
Look for Mirror Copies >= 1 and LV status = available/syncd. If both of these conditions are present for every logical volume then you
truly mirrored. I assume anyone who did this also took care of the mkboot stuff.

If you are mirrored then the replacement is really easy if not it could be an OS reload.

You can contact HP contracts at 1-800-386-1115 to see if you have a service agreement in force.
Good Luck
If it ain't broke, I can fix that.
A. Clay Stephenson
Acclaimed Contributor

Re: K-Class SCSI Critical Error

Tim,
I forgot to mention if this is the boot disk and especially if it's not mirrored, ckeck out the make_recovery command (part of Ignite-UX).
If it ain't broke, I can fix that.
paul courry
Honored Contributor

Re: K-Class SCSI Critical Error

It's your boot drive.

10/0.6.0 is the default address on the K460 for the boot drive. 10/0 says it is internal; 6.0 is the standard SCSI address for the boot disk (2.0 is the CD-ROM and 0.0 is the tape drive)

You don't happend to have a DLT drive on this bus do you? That happens to be a known cause of SCSI errors. Also, have you used XSTM to strip the drive and look at the error logs? This might give you some insight on whether this is a reoccuring phenomenon or a one time thing.
Tim Krego_1
Frequent Advisor

Re: K-Class SCSI Critical Error

Paul,

There is a Quantum DLT connected to 10/0.4.0 that has been hooked up since Feb.

I do not know what XSTM is so I will have to look for that.

This is the only error in syslog.log since a reboot on 3/14/01.

Thanks
Tim
HP/UX Newbie
A. Clay Stephenson
Acclaimed Contributor

Re: K-Class SCSI Critical Error

Tim, I glad to hear that you've only had one error logged since the boot. Did the error occur while the tape drive was active? Normally DLT's (and most especially 7000's & 8000's) should not be on the same bus with other devices. In fact, if it's anything but a 4000 you are really only supposed to have 1 device per scsi bus. Still if the root drive is showing errors I would definitely look into replacement soon unless the tape drive was possibly active or being queried at the time of your error.
If it ain't broke, I can fix that.
Tim Krego_1
Frequent Advisor

Re: K-Class SCSI Critical Error

A. Clay,

The tape drive is a DLT4000. I am not sure why it is hooked to that interface. I assume because we don't have another interface to hook it to.

The error happened at 2:00 EST yesterday so no backup was running at the time. The machine is at a remote location so I don't know if someone may have tripped over a cable, etc.

I use telnet to connect to the machine. I can't get Exceed to connect remotely, but that is another issue. I cannot run XSTM, but I did try to run MSTM but have some problems with the text display.

Since this is the first error I am hoping that it may be a fluke. HP support is supposed to call me tomorrow to purchase a contract. I am going to create another make_recovery tape now.

Thanks again
HP/UX Newbie
paul courry
Honored Contributor

Re: K-Class SCSI Critical Error

Strongly suggest you spend a few dollars and get that DLT 4000 off 10/0 and put it on it's own dedicated card. It may or may not solve the problem, but it should absolutely not be there. (Considering the number of late nights I've spent at several customer sites I have strong opinions on this)

To give you an example, two DLT 4000's shared a single SCSI card and backups were taking 12.5 hours. (KS989-650 3.75Gb RAM EMC disks) With the addition of another card dedicated to the DLT's the time dropped to 10.5 hours. Does this tell you something about the bandwidth saturation we were experiencing?
A. Clay Stephenson
Acclaimed Contributor

Re: K-Class SCSI Critical Error

Tim, I strongly agree with Paul. The fact that it is a DLT4000 is probably the only thing that saved you but it is still a no-no.
If it ain't broke, I can fix that.