Auto Req Help
17 messages in this thread
Help! request: Maybe someone here can help, I hope. I have been using a
patch program called CANCEL! to automatically, via the software, simulate a
"click" on the cancel button of auto-req. things like "disk xx has a
read-write error". It's been most necessary with the Syquest drives I'm
using (lots of disk errors while rendering and recording multi-frame
sequences).
The problem is : CANCEL! won't work with WB2. I have tried
"SAFEBOOT", but that doesn't cancel R/W errors. Can anyone suggest (or
write, $$ possible) a program to do the same? I think CANCEL! was written
by Bryce Nesbitt, but I don't know how to reach him to see if there's a 2.0
equivalent. Any help much appreciated.
—Jeff.
Jeff;
If you are getting lots of errors using your SyQuest then I would suggest
that you try to find out whats wrong with the setup. You must have
something seriously wrong there. I have had three bad carts but other than
those, I do not get errors on the SyQuest.
-bill/ARS KB7LX
I've lost 4 carts, and had similar problems with the r/w errors on two
different drives and systems, each of which have gone through 2 variations.
Maybe it's the cat hair? —jeff
could be (the cat hair)
-bill/ARS KB7LX
Since I talk to Bryce on a regular basis, I will give him a shout and ask
about it. I do think that you need to address the hardware errors first,
simply canceling the requester does not seem like a good thing to be doing.
It is a certain formula for losing a disk.
I used the "ignore this error" method for over a year with WB1.3, and
didn't encounter any other problems other than the same ones I'm having
now. It would be a short-term fix, anyway, since what I really want is a
300 or 600 meg drive so I can *go away* rather than babysit for a disk
swap! Thanks for any help, though: a short-term fix is better than none!
—jeff
I am absolutely amazed that you haven't had a problem. R/W errors are a good
indication of something seriously wrong. If it ever happens to fail to write
to the root block or the bitmap, the next time you go to access files, you
are in serious trouble of losing another file.
John,
Remember the 2090 and SCSI drives? You'd get R/W errors all the time
that had nothing at all to do with the media, but was an artifact of DMA
loading and the 2090 DMA controller. Canceling those was the normal
solution.
Dean
DKB Software
No, canceling them was the wrong solution. It got the error and did *NOT*
always complete the write (from a person who had a 2090/ST-506 drive and lost
it a number of times due to a program that did turn off requesters). The
appropriate response was to *RETRY* the failed operation.
John,
Ouch, your right. For some reason I was reading (and typing) that the
wrong way. RETRY was the correct solution, NOT CANCEL.
Dean
DKB Software
Dean,
Until just a bit over a year ago, when I got a 3000, I used a Wren III with
a 2090 on a 2000 (mb rev. 3.9). I _never_ had DMA contention.
-sja
Steve,
Were you using overscan? If not that is probably why you didn't have
problems. Even with overscan, not every machine seemed to have the same
difficulty. I know mine did with a Seagate ST277N.
Dean
DKB Software
Dean, Yes, I was running overscan and hires/interlace. I saw the slowdown
on other's machines but never experienced it on mine. I was running with a
2090, 2 ST-506 MiniScribes and the Wren III.
You must have been lucky. The problem was well known at Commodore. In
particular it showed up with hi-res overscan. Dpaint could make it happen
just about every time.
I know it was a well known problem. I saw my dealer's machine (2000 with
2090 and SCSI HD) take over 15 minutes to load PageStream. It just never
happened to me — and I used PageStream and DPaint a couple of times a
week.
John,
The problem was also fixed in their latest (although now…rather
old) driver. I'm still using a plain old 2090, with both SCSI and ST-506
drives and *never* get read/write errors from any of the drives. I do get
slowdowns when there's heavy DMA and I try to access one of the SCSI
drives…but no requesters.
Don
Actually, what they did was simply implement the retry mechanism in software
(although a bit more sophisticated than that).