StorageWorks disks disappear when OpenVMS is rebooted.

[Kirk Reindl]

AlphaServer AS1200
OpenVMS 7.3-2
Compaq Storage Works disk array
----connected via hsg80 fibre controllers

Hello,

I'm running into a strange phenomenon. When I reboot my standalone AS1200 all my devices on my storageworks fabric/SAN are missing after reboot.

If I then run mcr sysman io autoconfigure, all my devices come back. This happens everytime I reboot.

Here is what makes this problem even stranger. This AS1200 server is also connected to an EMC SAN using different HBAs. And the disk on the EMC Symetric are presented to the OS everytime after I reboot and I don't need to run io autoconfigure.

Any ideas on why the EMC disk gets configured on reboot, but the storageworks disk needs io auto to be ran.

I've checked operator.log and errorlog but no clues. There is no zoning on the storageworks SAN.

Any ideas.
Regards,
Kirk

[Uwe Zessin]

Strange thing.

- Does the AS1200 boot from a local disk or from the HSG?
- Does it access the HSG through cascaded FC switches or is it connected to the same switch as the HSG?
- What firmware revision is on the FC adapter in the AS1200?
- Is there a graphics card in the adapter? Which one?
- Is the BOOT_RESET flag enabled on the AS1200?

[Antoniov.]

Hi Kirk,
on old VAX there was the MSCP_SERVE_ALL parameter to start disk and HSC controllers; might be is valid for EMC SAN too?

Antonio Vigliotti

[Uwe Zessin]

Antonio,
MSCP_SERVE_ALL controls the behaviour of the host's local MSCP server. A fibre channel LUN appears as a locally attached device if it is accessed through a local fibre channel adapter.

[Ian Miller.]

Are the disks visable at console level? What are BOOT_RESET and PROBE* console environment variables set to?

[Uwe Zessin]

Watch out. If you shut down the system, then you need to INITIALIZE on most AlphaServers before you can use WWIDMGR. Well, this is an AS1200 - if I remember correctly you have to enter
>> SET MODE DIAG

before you can run WWIDMGR. This can falsify the results. But Ian's question made me think about another experiment:
- shutdown the system
- do an INITIALIZE
- boot and see if you can now see the disks

[Kirk Reindl]

The AS1200 boots from a local disk.
The FC switches are not cascaded and the fibre (HBA) is connected to the same switch as the HSG.

Firmware on the FC adap in question is:

SDA> fc show dev fgc
FGC0: operational firmware revision SS3.20X7
port_name(adapter_id) = 1000-0000-C921-547C, node_name(host_id) = 2000-0000-C921-547C
SDA>

The boot_reset flag is set to ON
I can't seem to find a probe console variable
I do a show probe, but nothing comes back.

None of the storage disks are available at at the P00>>> level at boot up.

I've swapped out the fibre cable. That didn't work. Then I tried a different port on the switch but got the same results.

Any other ideas/direction would be great.

Thanks,
Kirk

[Uwe Zessin · Accepted solution]

Right, the SRM console does not know how to deal with fibre channel devices directly. That's what the WWIDMGR I was talking about is for.

On the AS1200 you have to enter diagnostics mode first:

>>> SET MODE DIAG

Now you can look for the devices:

>>> wwidmgr -show wwid -full


Can you tell what type of fibre channel switch you have and the firmware level?

[Kirk Reindl]

Hello all,

Thanks for all the responses. A colleague and myself were continuing to troubleshoot and he noticed the port on the compaq fibre switch was set to "negotiate". Once he locked it down at 1Gbps, we were able to talk to the disk at the SRM level using wwidmgr commands. Most importantly the devices on the fabric are online at the OS level after reboot.

One noteable thing.

We have a different style alpha (ES40 running 7.3-1) that doesn't experience this problem. This server has fibre cards that get connected to the switch and they are set to "negotiate". However the style of fibre card is slightly different in the ES40.

ES40 has KGPSA-C cards
1200 has KGPSA-B cards

Regards,
Kirk

[Uwe Zessin]

Stupid me - I had overlooked your comment that the server was connected to two different fabrics, so I was wondering why oh why the system could configure one array and not the other. Sigh :-(
But thank you for the update.

Well, about this autonegotiation thing...
There can be different behaviour depending on firmware revison. I once had an adapter that got a link when the server was booted, but the link failed to come back when the cable was pulled from the switch and then connected back. A firmware update fixed that problem.

I have checked and as far as I can tell the latest revision for the KGPSA-BC adapter is indeed 3.20x7.

In case you want to put a DS-KGPSA-CA into the AS1200, please note that you might see problems with firmware revision 3.91a1 - use 3.82a1 instead.