Home > Computer network, Linux > HP-UX problem with disks from SAN

HP-UX problem with disks from SAN

We had project replacing SAN switches. Customer has one old HP-UX (HP Unix) system. This system is very sensible about SAN changes.

Normal situation

You can see all hardware using command (http://www.bga.org/~lessem/psyc5112/usail/man/hpux/ioscan.1.html):

ioscan -fnC

When you change anyting on paths for devices you will need to re-initiate those devices with following command (http://www.bga.org/~lessem/psyc5112/usail/man/hpux/insf.1.html):

insf -e

This command will reinstall drivers for all devices and will discover new paths to SAN devices and you will see new devices. When you look into devices after re-initiating drivers you can see hardware with status NO_HW using command ioscan -fnC. Those are devices which are not actual now.

You need to remove those devices using script, which uses command rmsf (http://nixdoc.net/man-pages/HP-UX/man1/rmsf.1m.html):

`ioscan -fnC disk | grep NO_HW | awk ‘{print “rmsf -H ” $3}’`

Problem

But we, again, had problem with this system. When we re-initialized drivers, we had only old paths and NO_HW (dead) paths to SAN devices. We didn’t have new paths which should be rediscovered during re-initialization. So now started hunt for solution and root cause. After some time someone tried to look into system logs using dmesg comand and we saw following error:

Configure the switch to assign a domain value other than 8. Domains > 8 will not be recognized till this problem is fixed. td_create_domain_node: hw_path = . For backward compatability domain 8 is not allowed.

So now we had solution and also root cause. HP-UX knows domain ID from SAN switches and it has reserved domain id value 8 for Private Loop devices. So do NOT use domain id 8 when you have HP-UX connected in your SAN.

We spent about 2 hours looking for a solution and I hope it helps some one and saves him some time 🙂

  1. No comments yet.
  1. No trackbacks yet.