Tape device polling over the SAN other than dm_stape

[Mike Johnston]

I am looking for documentation that will help me identify what processes are polling my SAN tape drives. (I have confirmed dm_stape has been removed from the STM configuration)

In my NSR trace logs I see what I have been told are poll events

PS_DrvSend: Bus:0 Target:2 Lun:0 CDB:x12

SCSI msg out c0010203 cdb 12018000ff0000000000000000000000

FrmHdr 01623613 00622113 08900008 00000000 805d006c 00000000

[Ralf Loehmann_2]

Hello Mike,

I already answered your other question about the NSR. What I see here is a standard inquiry, requesting page 00. This should not be any problem for the NSR and it is for sure not the reason for any problem.
This is a fully supported polling to check if that unit is alive.
I do not like the polling, but this should be handled by the bridge and should not result in any problem.

Recommandations:
get NSR to latest firmware
bring HP-UX patches to latest version on all hosts in the SAN (patches for SCSI, stape, fcms, tachlite td)

Try to catch a trace right at the time of a problem. This might be not easy and you might need to get a fiberchannel analyser to really get closer to the problem, but before that I would follow the HP recommandations (get the SAN Checklist).

Ralf Loehmann.

[Mike Johnston]

The reason I am pursuing this line of thought is: I am receiving "Cannot write to device ([5] I/O error)" errors on devices that have writtien hundreds of Gbyts of data then suddenly become unavailable. I believe we are SAN compliant (ats_enabled=0, san_safe=1, dm_stape disabled, latest DP 5.1 patches, acceptable UX patch bundle). I have read in KMINE snipets of information about "bad" polling other than dm_stape but I cannot find any details or documentation on how to resolve.
Any suggestions on that line ?