geoff@utstat.uucp (Geoff Collyer) (12/18/88)
We have just had a Swallow fail, about two weeks after installation. The symptom is that about 15 minutes after cycling power on the drive, the Swallow begins to complain spuriously of swap errors. Cycling the drive's power will cure the condition for about 15 minutes. No data is lost by all of this. Has anybody else seen this or a similar problem? I used to have great faith in Fujitsu, modulo the SuperEagle disaster. -- Geoff Collyer utzoo!utstat!geoff, geoff@utstat.toronto.edu gets(3) must go; this is not negotiable.
dennis@gpu.utcs.toronto.edu (Dennis Ferguson) (12/18/88)
I don't know if it's related, but we had a hell of a time when we first installed Swallows on a Ciprico Rimfire controller. We eventually deduced that the disk was getting scribbled on if we powered down the CPU with the disk readied. The error messages most often said something about "Sector Not Found". It required a reformat to correct the trouble. Frequently the cylinder with the disk label would get taken out, though a couple of times something further back on the disk was damaged. After reformatting parts of the disk five or six times, we started to get the idea and began to religiously write protect the disk whenever we were going to power down the CPU. We haven't had any problems since. Actually, write protecting the disk when powering off the controller is probably a good idea in any case (maybe this is why I've never heard anyone else complain about this?), but we'd never owned a disk/controller combination which didn't behave benignly when doing this and we'd gotten lazy as a consequence. Dennis Ferguson University of Toronto
daveb@geaclib.UUCP (David Collier-Brown) (12/21/88)
From article <1988Dec18.150601.8808@gpu.utcs.toronto.edu>, by dennis@gpu.utcs.toronto.edu (Dennis Ferguson): > [...] We eventually deduced that the disk was getting > scribbled on if we powered down the CPU with the disk readied. > The error messages most often said something about "Sector Not > Found". Hmmn... I wonder if this applies to other controller/powerdown situations. We once had a site with repeated power outages, leading to disappearing sectors/records. Repeated as in "at lunch, both coffie breaks and 5pm". I can well understand spikes causing strange patterns appearing in magnetic media, but I wonder about anti-spikes... Comments, hardware-people? --dave -- David Collier-Brown. | yunexus!lethe!dave Interleaf Canada Inc. | 1550 Enterprise Rd. | He's so smart he's dumb. Mississauga, Ontario | --Joyce C-B
duncan@me.utoronto.ca (Duncan Poole) (12/23/88)
It only takes one line to change state with the E-SMD interface to indicate a write. Seems to me on powering up a controller board it might be easy to have that happen. Older and wiser souls have mentioned that in the "old days" one religiously write protected all disks before a power cycle. We certainly do that now, there's a nice secure feeling when you follow that procedure. Dunc.