Conference ssag::ask_ssag

    The error log indicates that the disk drive is rejecting a
    Mode Sense command for page 0.           
    
    In general, 3rd party disks are not supported.  Some of them work,
    some of them don't.  Digital does not guarantee that VMS V6.2 will
    work with every disk drive on the market.

    Thanks,
    
    SCSI_INFO indicated, that the drive has problems with mode sense 10
    commands!
    
    Volker

    Since you mention this was on an Alpha OpenVMS V6.2 system, you might
    try installing the latest V6.2 patch for Dkdriver;  some fixes were
    added to accommodate third party drives in the area of mode page
    handling after V6.2 shipped.
    
    The error log entry you posted would indicate that the device returned
    a bad SCSI status after a mode_sense_6 command.  It is not likely a
    mode_sense_10 problem, because in V6.2, Dkdriver will back off
    transparently to a mode_sense_6 command if the initial attempt at
    mode_sense_10 returns a failure.  So it was the mode_sense_6 that
    likely failed, although it's not clear on what specific page;  by the
    time the error is logged, the command itself is long gone, so what
    the 1a000000006400 represents in the error log is just a token entry
    to mark the the event (doesn't implicate page 0 -- Dkdriver doesn't
    query page 0 anyhow). 
    
    Sue

>It is not likely a mode_sense_10 problem, because in V6.2, Dkdriver will back
>off transparently to a mode_sense_6 command if the initial attempt at
>mode_sense_10 returns a failure. 

Correct but ONLY if the device returns a failure to the MODESENSE(10), however
some device do not return a failure, they return good status with bad data. VMS
then takes this bad data and converts into bad logsense info. This then gives
you "Invalid Modesense Data" errorlog entries and causes the FIRST mount after a
boot to fail.

This is what happened to the RZ25M, RZ25L and RZ28B's running their original
versions of firmware once 6.2 was released.

roger.

    Re:  .4   Actually, I believe that the good-status-but-bad-data problem 
    was discovered during V6.2 field test and fixed by the final release,
    so this would seem to be a different problem.  At any rate, a Dkdriver 
    patch still seems worth a try.
    
    Sue

    I ran SCSI_INFO on an RZ28B which was logging "invalid mode sense"
    errors, and crashed the customer's system! 
    
    The bugcheck was "inconsistent IO database entry".
    
    OpenVMS 6.2 with ALPSCSI02_070 installed.
    
    The last output before the system went down was in SCSI_INFO:-
    .
    .
    .
    <<$! Issuing 10-byte MODE SENSE QIOW to get current values for page01h>>
    .
    .
    .
    CRASH!
    
    
    	I'm waiting for a parcel of brown smelly stuff from the customer to
    arrive on my desk!
    
    	Any offeres of explanation, solutions, or armour gladly recieved!
    
    				Dave Clark
                                      

   Acquire the CLUE output from the crash and send it along to the CANASTA
   e-mail crashdump server.  (See VMSNOTES 223.* for CANASTA info.) 

   Then -- if you tire of looking at the dump and if CANASTA does not
   recognize this crash -- forward the dump along to OpenVMS engineering.

   (I doubt that the SCSI_INFO program is itself directly responsible for
   the crash -- though it appears to have provoked the crash -- and this I
   would expect to find an error in a SCSI-related device driver.)

RZ28B's running 003 firmware do not like modesense(10) commands with a greater
than FF(x) length. That ALPSCSI kit you installed was supposed to change it so
that mount's would use modesense(6). Get the vms folk involved.

roger.

    re: .7 & .8
    
    	Thanks for the feedback. 
    
    	I am in touch with the site engineer in connection with upgrading
    the RZ28B firmware to '0006'. I'll get him to copy the crashdump to
    tape and route it in to me where I can make it available on the
    network.
    
    	Watch this space!
    
    				Dave