Looking at "armdsp -t" output

[Denver Osborn]

My VA7100 is in a warning state,

WARNING: The configuration of the subsystem is such that one or more drives has only a single path of access to it.


So I looked at "armdsp -t" and see that disk D1 seems to be having issues. Then I noticed D13 looks like it's only being accessed via one port as well. The amber light on D1 is flashing, so I'll be replacing it. D13 is is good except for what I noticed in armdsp.

Please have a look at the attached armdsp -t output and offer suggestions.

Thanks!
-denver

[goldboy]

Denver,
I am wondering what is the armdsp -a results ?

I had a customer that had the same issue but beside a drive that went bad he had lost a controller.
The controller did not have any amber lights on and was displaying the disk information but in logprn results it was reporting errors.

run the logprn on the past 2 days and also armdsp -a and send me the results.
Tal

[Denver Osborn]

Hi Tal, thanks for the feedback on this. It never hurts to have a 2nd set of eyes.

D1 wasn't seen by the array on Friday afternoon. Then on Monday morning D13 was reseated and the array took it and formated D13 . Then during the hotplug event D1 came back online but with only one port. The logprn is from Friday to Monday morning. Seeing a lot of backend scsi events from C2.

It might also be worth noting that M/C1.H1 was set to address 108 and M/C2.H1 was 109. Not sure why, so I set them to 110 and 108.

-denver

[goldboy]

Device Unique Name = 50060B0000092972
Device Name = HP StorageWorks Virtual Array 7100
Date = March 16, 2007
Time = 10:00:13 AM CST
Controller Time Stamp = 2192038615374
Event Number = 8871
Sequence Number = 200
Event Type = Controller Event
Event Code = 30
Severity = 2
Event Count = 3
FRU Location = M/D1
FRU Data = enclosureId/slot/component/subcomponent : 0x00/0x10/0xff/0xff
Reporting Controller = M/C2
Event Source Info = serialNum/moduleId/processId : 00PR03010114/0x4/0xffffffff
Event Description = DEV_STATE_UNKNOWN_EH The I/O executed concurrently with a device reconfiguration. The state of the drive was unknown (danger of write cache being enabled.) The I/O was not retryable or the PDD's retries are exhausted. The I/Oshould be retried by the client.
Event Unique Info = none



Denver,
As always huge log but most of the messages are to be ignored.
Did you try to re-seat D1 disk ? or to replace it ?

Tal

[Denver Osborn]

Actually tried both reseating and replacing.

Reseated the drive didn't help. So I manually failed it on Monday, then after the rebuild I inserted a new disk. The new drive was seen by the array as having 512-block size, but instead of formatting the new drive it went to "Init failed". Same thing happened with yet another new disk.

As you can see w/ armdsp output, auto format and include are on.

I've logged a case with HP, but it doesn't seem to be making much progress.

-denver

[goldboy · Accepted solution]

I assume they will probably replace the mid-plane the slot might have turned bad (cant point to a specidfic controller problem since both reported the multiple errors).

Hope they resolve it for you.

Tal

[Denver Osborn]

I performed a partial reset on my VA7100 and it seems to have cleared the errors and allowed me to replace D1 and it's formatting now.

I'll be following up with HP if the backend scsi errors start again.