From: O'Brien, Pat (pobrien@mitidata.com)
Date: Wed Oct 30 2002 - 11:56:49 EST
I recieved only 1 reply from the venerable DR Blinn,
I believe there is some bug in the kernel that breaks this. I am now
the (unfortunate) DRI for the "hwmgr" command and the kernel code
that it invokes directly; the "scp" is in fact "owned" by the mass
storage subsystem code, and there are cases where that code comes
to believe that it should NOT let you delete certain objects in the
database.
If it were my system, I might try this: Shut down to single user
mode (if it's a cluster, shut all the members save one all the way
down, then shut the remaining member down to single user mode).
Use the "bcheckrc" script to remount the local file systems. Then
try using the "hwmgr" commands. If they STILL fail, then there is
something going on that the SCSI subsystem needs to fix. I don't
know how this "bug" got in, but I'm pretty confident (at this point)
that there is a bug in this area; I've seen reports of a similar
problem from one of the qual labs.
Tom
Original post
I have 2 hsg80's which have been redeployed from this tru64 5.1 system to
another site. I have removed the device special file for the luns which
were removed. I be damned it I can remove the command console ports scp. I
have tried hwmgr -delete scsi -did # and hwmgr -delete component -id # to no
avail. PLease show me the errors of my ways!
pmob
This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:48:58 EDT