StoreEver Tape Storage

cabling msl5052 4 drive to old Compaq MDR

 
SOLVED
Go to solution
Vincon Daniele
Frequent Advisor

Re: cabling msl5052 4 drive to old Compaq MDR

Marino wrote: could you collect a full report....

I Marino, what type of report you need? trace log and scsi command?

In monday i visit my customer and i send all trace log.

Thank's
Marino Meloni_1
Honored Contributor

Re: cabling msl5052 4 drive to old Compaq MDR

I'm seeing several HW errors on the scsi bus, related to both drives, (diconnections,bus reset and write errors) these do not seems strictly related to the assestion, but need to be investigate.
I also see that more than one host is issuing commant to the drives. Have you a zone just for backup in your san? have you several servers that can see the library? in this case you need to apply rules to all the servers (TUR disabled, Agent 7.2 or higher with FC agent disabled, RSM stopped....

For the assertion, I would be nice to have the complete report, (not only the log) including configuration and assert traces.

answering your last question: everything in the report


Vincon Daniele
Frequent Advisor

Re: cabling msl5052 4 drive to old Compaq MDR

Goodmorning ! in attachment all trace and report of my customer.

We have 2 SAN, one at 1 Gb with MA8000 and other at 2Gb with EVA3000.
But the problem is the same, Assert error and random reboot of NSR.

In all SAN are Zoning (port zoning) and the firmware & Drivers are update.

Thank's
Marino Meloni_1
Honored Contributor

Re: cabling msl5052 4 drive to old Compaq MDR

Could you tell me what HBA you have?
and if it is a emulex or KPGSA, have you the parameter ResetTPRLO set equal 2 in the registery parameter?
Vincon Daniele
Frequent Advisor

Re: cabling msl5052 4 drive to old Compaq MDR

Yes the parameter ResetTPRLO is set to 2 in every HBA ,
On 1Gb SAN we have LP8000
On 2 Gb SAN we have LP952

Thank's
David Ruska
Honored Contributor

Re: cabling msl5052 4 drive to old Compaq MDR

Vincon,

Marino did an excellent job reviewing the event log.

I spoke with the division engineering team to get their comments on the event log as well.

First, as Marino mentions, there are multiple host IDs attempting to communicate to the library, most likey doing some type of polling. You can see log sense commands in particular. The host(s) sending these commands are most likely using a very short timeout, and then attempt to abort the command and eventually sends a bus reset. This is bad news when the library controller is on a shared bus with a drive - it could kill an I/O that is pending with the drive.

So I would suggest that you look for and disable any polling applications that could be communicating to the library, or alternately change the zoning and/or NSR mapping so only the host controlling the robot has access to the library controller lun.

Second, there are multiple DSA errors after a write error, and eventually an assertion.

The sequence matches a problem we've identified as a possible issue in 5.6.69 (and likely earlier) firmware and are working on a fix. If you can open an elevation with HP support, and have it raised to tier 3 we can provide you with the fix when it is available.

The workaround for this issue is to eliminate the cause of the write errors (e.g. by filtering media, running drive assessment tests, etc).
The journey IS the reward.
Marino Meloni_1
Honored Contributor

Re: cabling msl5052 4 drive to old Compaq MDR

I reviewed the traces, and cannot found a root cause on it, the traces are a very shoort snapshot about the activity, and it is not easy to capture what is needed.
Anyway I would start the troubleshooting perfoming the suggested list of actions, this should start to give you some results, and I my opinion is better than trying to install a MDR

marino
Marino Meloni_1
Honored Contributor

Re: cabling msl5052 4 drive to old Compaq MDR

Also, it may be time to call HP support ;)
David Ruska
Honored Contributor

Re: cabling msl5052 4 drive to old Compaq MDR

Just to be clear, this sequence in the trace:

62. 08/12/2005 21:13:25 2d04h40m35.12s SCSI port 1 invalid DSA
63. 08/12/2005 21:13:25 2d04h40m35.12s SCSI port 1 invalid DSA
64. 08/14/2005 04:25:04 3d11h52m14.75s Assertion failed: 0, file ../ps_driver.c, line 3119
65. 08/14/2005 04:25:04 3d11h52m14.75s Assertion failed (continued): 3119

It does match an issue we (HP engineering) are working on and should have a solution available soon. It is an error condition that's triggered by a write error, so reducing those is a workaround.

Please open a case with HP if need early access to the fix for the above.
The journey IS the reward.
Vincon Daniele
Frequent Advisor

Re: cabling msl5052 4 drive to old Compaq MDR

David Ruska:
Please open a case with HP if need early access to the fix for the above.


I'm in contact with HP support people and we are investigate in all components of SAN
I inform you if resolve the problem.

Thank's