What To Do When You Start Getting SCSI Disk Errors

One of our servers here at ONE/Northwest (which, among other duties, hosts this blog!) has been acting a little flaky over the past few weeks.  The SCSI disk subsystem was throwing intermittent clusters of CRC errors, which seemed to be causing the filesystem to go into read-only mode to protect itself.  Which would of course bring down pretty much all of its processes, requiring a restart and an fsck disk check.  Not good at all.

After Googling around a bit, and pondering the hassles of SCSI driver upgrades, I decided to just get a can of compressed air, disconnect all the drive cables, blow everything out, then carefully reconnect everything.

Twenty-four hours later, and not a single error.  Another day or two and I’ll be satisfied.

It’s nice when the easy fix works.