[comp.periphs.scsi] Tar 150 MB 1/4" cartridge read problem.

masaru@media-lab.media.mit.edu.MEDIA.MIT.EDU (Masaru Sugai) (05/12/91)

In article <1991May10.211712.5745@hal.com> rana@hal.com (Marty Rana) writes:
>	I will appreciate receiving helpful hints on a problem reading
>a tar 150 MB 1/4" cartridge on archive tape drive (QIC150 compatible).
>It successful reads about half of the 14 MB file on the tape and
>then gives Read I/O error. Same is the case with 'dd' command with
>various blocking sizes. This tape was written on a Tandberg 150 MB tape
>drive. The Tandberg drive successfully reads the tape written using it.

I had the similar problem when I hooked up Commodore's QIC drive (CALIPER
150A) to NeXTstation last week. My drive worked fine on Amiga at the transfer
rate around 50KB/sec. On the other hand, NeXT's generic SCSI tape driver (ST)
showed an unusual behaviour. WRITE is quite slow (say, 10KB/sec), but works
fine. On the other hand, READ is nearly close to the specs (110KB/sec), but
it always fails on large tar files, over 50MB.

I noticed that all errors occurs when drive changes its winding direction 
at the end of track. (FYI,  DC6150 has 18 tracks, and each track records
up to 8.4MB.)  Error messages says "RESELECTION TIMEOUT" on console.

For a quick remedy, I wrote a tape reader which changes timeout to a larger
duration, and retries the same command several times on errors. It seems 
to have get over my problems, but I'm still hesitant to take a critical backup.

I wonder there is something incompatible with SCSI chips on both ends. Are 
there any tips to write reliable SCSI program for non-disk devices whose
mechanical delay needs special attention ?  Thanks in advance.
-- 
-- Masaru Sugai:Use disclaimer. CIS 72050,2141:NeXT + A3000 = money-eater
NEC Corporation:sugai@ccs.mt.nec.co.jp DORMANT:hardwired logic,machine language
MIT R.Affiliate:masaru@media-lab.media.mit.edu:  "Silicon on Sapphire" by CLASH