MSA 1500 logical disk (in)visibility on a server

[Modris Bremze]

Hi.
We have an MSA 1500CS on one side and an HP-UX 11.23 ia64 on the other side (which is also user to manage MSA through hpacucli).

There already was one logical disk created in MSA, which was presented (unmasked) to the server and being successfully discovered and used. The weird stuff started when I tried to present another logical disk to that server.

Masking/unmasking happened without errors in hpacucli, but the disk was not visible on the server (was not detected with ioscan -fnC). I then tried to delete the logical disk and create/present a new logical disk - same result. The I deleted the freshly created disk and created a new one, but smaller this time (figured that maybe 1.9TiB before was too big or something). Again - no joy. Then I tried to present another disk in the same array on MSA and that disk was detected.

And then I sort of ran out of ieas. Maybe someone else knows what could be wrong here?

[Torsten.]

SSP used and configured? Better to check this again.

The most important thing is to use "HP-UX" presentation mode, anything else will produce trouble.

Is the MSA already active/active?

[Modris Bremze]

This particular MSA is being used for over a year now. SSP is enabled, we have other logical disks made available to other servers with SSP. After "ld [n] modify unmask=[wwn/connectionname]" the "show ssp" command and "ld [n] show" displays correct information - the server port WWNs are right. What exactly do You mean by ""HP-UX" presentation mode"? If that is the same as "Host mode", then it is displayed as "Default" for all adapter IDs and, as I said, doesn't make other disks not detectable.

[Torsten. · Accepted solution]

Default is wrong - must be "HP" for HP-UX.

Otherwise you are in trouble.

[Modris Bremze]

Tried that, and it actually solved the problem. Thanks, Torsten.
I should mention that by changing Host mode of an Adapter ID, the H/W path of the presented disks changes on that server, so one should be careful if there are other logical disks already unmasked/presented on that Adapter ID, as did I. So I also had to do a vgexport prior to changing the Host mode in MSA and vgimport afterwards. Thanks again, problem solved.

[Torsten.]

Good news.

With wrong host mode you can see strange things, e.g. 1 LUN with a loop connection, same LUN with a fabric connection (different path, different device, same LUN) etc.

So the changed path is expected, more or less.
I should tell you, right? Sorry.

AFAIR with wrong host mode you can sometimes see LUNs up to LUN 6 or 7, but not a higher number.


However, wrong mode = trouble.

Have fun!

[Modris Bremze]

Yes, wrong/default host mode can produce weird results even if everything was working before. Now I know that too.
And no worries about not mentioning the path name change, everything went fine with vgexport/vgimport. Thanks.

[Modris Bremze]

.