[2087] in SIPB-AFS-requests
Plan for dealing with rosebud
daemon@ATHENA.MIT.EDU (ghudson@MIT.EDU)
Fri Aug 4 04:11:13 1995
From: ghudson@MIT.EDU
Date: Fri, 4 Aug 1995 04:10:20 -0400
To: sipb-afsreq@MIT.EDU
Here's what I plan to do about rosebud. Tom seemed to think this was
reasonable when I described most of it to him.
1. Replace all the SCSI cables and terminator. This will
cause a ~15-minute outage. It's unlikely to solve the
problem, but the next step is going to take more effort
and involve more outages, and this will at least tell us
it's not a simple cabling problem.
2. If writes to the 4GB Seagate disk are still slow, remove
all the data from the Seagate disk (back to the Micropolis
disk), take down rosebud, perform some read and write
timings to the disk from rosebud, and then do the same
tests on yaz. If disk accesses are slow on rosebud and not
on yaz, then we will have narrowed down the problem to the
host adapter. If disk accesses are slow on both, then we
send the Seagate disk back.
3. If it's a problem with the Seagate disk (not the SCSI bus),
then we'll ask for an RMA on the Micropolis disk at the
same time, since we will still have reason to believe that
there's a problem with that disk.
