home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.sys.sun.hardware:5505 comp.periphs:1507
- Path: sparky!uunet!charon.amdahl.com!pacbell.com!ames!sun-barr!cs.utexas.edu!uwm.edu!ogicse!das-news.harvard.edu!husc-news.harvard.edu!burrhus!ddl
- From: ddl@burrhus.harvard.edu (Dan Lanciani)
- Newsgroups: comp.sys.sun.hardware,comp.periphs
- Subject: interesting tape interaction
- Message-ID: <1992Nov12.013140.21432@burrhus.harvard.edu>
- Date: 12 Nov 92 01:31:40 GMT
- Article-I.D.: burrhus.1992Nov12.013140.21432
- Followup-To: comp.sys.sun.hardware
- Organization: Harvard University, Cambridge, MA
- Lines: 38
-
-
- I observed an interesting problem the other day and, though
- it may be related to partial faults in some of the components involved
- (and while it certainly depends on a specific combination of components),
- a description might possibly save somebody some time. My configuration
- consists of a Sun 3/260 with a Xylogics 472 controlling a CDC (Keystone?)
- 9-track 1600/6250 bpi drive using the buffered Pertec interface running
- at 500kBps.
- Having a file which was just a little too large to fit on one tape
- at 1600 bpi with a 20k record size, I decided to try a 40k block. The write
- operation failed with (according to the error code) a data-late condition.
- This didn't seem totally unreasonable since I don't know the 472's sustainable
- data rate and 500kBps might be marginal depending on the exact buffering/DMA
- interactions.
- I decided to check into this in detail later but switch to 6250bpi
- for the current tape. However, upon starting the copy, I noticed that the
- drive was doing an unreasonable amount of repositioning and remained near
- BOT. Soon the Sun timed out and aborted the copy, but the drive continued
- to thrash. Eventually the drive stopped but the Sun was clearly confused
- and would not allow further access to the device. Thinking that the
- driver might have been confused from the original data-late condition,
- I power-cycled everything and started over. Same result. I repeated
- this a couple of times.
- At this point I was beginning to worry that something had happened
- to the drive or controller and I ran some diagnostics on the drive. All
- seemed fine. And after the diagnostics (which, important point, write
- on the tape) the copy from the Sun worked. To make a long story short
- (yes, I know, too late) the CDC drive behaves very oddly when trying
- to write at 6250bpi to a tape with a long 1600bpi record size. I don't
- know if the process would eventually succeed, because the Sun gives up
- too soon. I do know that first overwriting with a short 1600bpi record
- avoids the problem. I also note that the thrashing looks suspiciously
- like the drive's normal auto-density-on-read-determination sequence,
- even flashing the density light once or twice. Why it would be doing
- this on a write-from-BOT is beyond me...
-
- Dan Lanciani
- ddl@harvard.*
-