topic Re: failover test - powerpath - failure - why ? in Operating System - HP-UX

failover test - powerpath - failure - why ?

INCS Dept. — Fri, 07 Sep 2007 04:49:29 GMT

Hi all,

At the moment we are trying to failover a node in our onvironment. Both nodes are not a member of a cluser; both nodes are (almost) identical if you look at the hardware; one nodes is a dev/test server, the other one is accp/prod server. Both servers are located at separate sites (geographically separated).

Twice a year we are performing a failover test; this means that we use the mirrorred LUN from our CX500 (one in each location) that and make these available to the server on the other location.

When we are trying to boot from the SAN the failover node can find the kernel; it has access to the LUN, i.e. all configurations that need to be made in order to make a failover test possible are ok.
When it find the kernel it straps to the moment when volume groups are being activated. Then it goed wrong; the server crashes and reboots because it cannot continue.

What am I missing ? The hardware is identical, and the kernel is an exact copy like all other data.

What is going wrong ?

Bye,

INCS

Re: failover test - powerpath - failure - why ?

Denver Osborn — Fri, 07 Sep 2007 06:03:12 GMT

When the server panics, what's the panic string? Any error messages might help track down the problem.

Also, have you checked that the LUNs aren't write disabled?

-denver

Re: failover test - powerpath - failure - why ?

INCS Dept. — Fri, 07 Sep 2007 06:21:13 GMT

I cannot give you a string; the server is up and running again. Furthermore, the message is only a split second on screen before it reboots.

Yep, the LUN is writeable. The default SP owns the LUN.

Thx.

Re: failover test - powerpath - failure - why ?

Wim Rombauts — Fri, 07 Sep 2007 06:22:57 GMT

Since when is it failing and what has been changed since it last worked ?
THe answers to that my be half the resolution.
Maybe your /stand/bootconf is incomplete
Maybe your lvlnboot configuration is incomplete

Re: failover test - powerpath - failure - why ?

Steven E. Protter — Fri, 07 Sep 2007 06:32:55 GMT

Shalom,

You are going to need to run another test and collect some data.

Also, I think its a bad idea to boot from the SAN. Yes, the system can do it, but when there are SAN problems you have virtually no system with which to diagnose the problem.

I suspect this is a configuration problem with powerpath, or SAN configuration.

SEP

Re: failover test - powerpath - failure - why ?

INCS Dept. — Fri, 07 Sep 2007 07:04:08 GMT

Yes, another test is being scheduled. And I agree that it's most likely that powerpath is the problem.

Maybe it was just an obvious issue; something I missed.

Booting from SAN; yes, it's a pain in the ..... Next month I'll transfer everything back to the internal disks.

Thx.

Re: failover test - powerpath - failure - why ?

Denver Osborn — Fri, 07 Sep 2007 08:22:46 GMT

Depending on the system type, you can view the console logs from the MP or GSP.

-denver

Re: failover test - powerpath - failure - why ?

Rita C Workman — Fri, 07 Sep 2007 09:36:58 GMT

Let's see if I get this....

Two servers - virtually identical, one at each site.
Two sites - geographically seperated

Two arrays - CX500 (Clariion)..? not sure based on your description.

What method do you you to replicate your data across to the other CX500 array ?

Let me know,
Rita