Conference ssdevo::hsz40_product

825.0. "Crash at SYS$DKDRIVER+D818 while patch HSOF" by TIKVAH::ARTHUR (Lampert, Israel Software Support) Mon Mar 31 1997 05:11

I just submitted the following to HAN::HALLE for inclusion in the Canasta 
data base. I am posting it here in the hope that it might help someone else.
Also, if disks should be dismounted before installing a HSOF patch, could 
someone please make that clear in the patch documentation (.TXT file)?

"I just submitted a VMS_AXP crash (CANASTA parameters only). Would you please
see that the rule is updated correctly? I do not have the dump, or even the
full CANASTA output; the site is restricted, and cannot release them.

The crash is at SYS$DKDRIVER+D818, failing instruction STL R11,#X0004(R0). At
the time of the crash, the customer was updating an HSZ50 running HSOF 5.0 with
patch 2. The crash occurred when the customer rebooted the HSZ50 after
installing the patch.

The HSZ50 was attached to the shared SCSI bus of a SCSI cluster. Both Alpha
systems were running OpenVMS V6.2. Only one system crashed. There are several
other systems in the cluster, including VAXes running V6.1. Obviously, they
are not attached to the shared SCSI.

The user documentation for the patch does not mention that the disks should be
dismounted before the patch is applied. The Blitz for the
MAXIMUM_CACHED_TRANSFER_SIZE integrity problem ("[TD 2250] Possible data
integrity problem with HSOF V3/5 - BLITZ") says to dismount all units before
installing the patch. The Blitz does not mention the patch number - it might be
3.

Without further evidence, I concluded that the most likely cause of the crash
was having the disks mounted while rebooting the HSZ50 after installing patch
2."

Arthur