Checking Disk for errors

[Ben_219]

I am running NFS server and sometimes it stops responding and I am suspecting it might be the disk. Can anyone how to check if one of the disk has bad sector?

I am running HP-UX 11.00

Any help is greatly appreciated.

[A. Clay Stephenson]

Normally, there will be error messages logged in /var/adm/syslog/syslog.log of the NFS server. I would say that network problems are much more common than disk errors for NFS.

In any event, a good disk test is to read the disk to /dev/null.

dd if=/dev/rdsk/c1t6d0 of=/dev/null bs=64k

This will take quite a while depending upon the size of the disk. It's rather unusual in HP-UX to use disk drives directly. Generally, LVM or VxVM is used to manage a group of disks.

I would look for and apply the latest NFS/Network patches.

[Ben_219]

I am attaching the syslog.log error regarding to EMS. I did not found any nfs errors.

[Prashanth.D.S]

Hi Ben,

Attached log cannot give much info on the reported issue.

Can you attach the below mentioned logs.

#/var/opt/resmon/log/event.log
#/var/adm/syslog/syslog.log

As mentioned in the earlier reply you can also run the dd command to check the disk status.

Best Regards,
Prashanth

[Sandman!]

I though you suspected a bad disk. You EMS log clearly reports a problem with the disk. Did you run the command suggested in the EMS log for further troubleshooting?

[Ben_219]

I already tried the commands in the EMS log, but unable to get the same result. However, I looked in the resmon log and get the following error (See attachment).

[Ben_219]

I already tried the commands in the EMS log, but unable to get the same result. However, I looked in the resmon log and get the following error (See attachment).

dd command did not produce any error. I think it is because it is just reading while my error occured during writting.

[Sandman!]

According to the resmon log your disk on I/O path 0/3/0/0.14.0 is bad. Check out the URL below and refer to "Event 100137"

http://docs.hp.com/hpux/content/hardware/ems/scsi.htm#100137

Media replacement is the suggested course of action.

~cheers

[A. Clay Stephenson · Accepted solution]

It does appear that you have a troublesome disk and syslog should have some errors in it on the NFS server.

Do this:

ioscan -C disk -fn and note the raw character device node associated with hardware path 0/3/0/0.14.0.

Then substitute that device for the if=/dev/rdsk/cXtYdZ device in my earlier dd example. That will low-level read the entire disk and should spot any errors. If you see no errors then lather, rinse, and repeat a couple of times. Bear in mind that disk problems can be caused by poor termination, bad/loose cabling, failing controllers, as well as failing disks.

[Ben_219]

I replaced the disk. It works fine now.

Thanks everyone for your help.