scsi tape curiousity

David Bear David.Bear at
Tue May 27 21:12:52 PDT 2003

I have two machine with scsi hard drives and tape units in them.
Since these were machine assemble 'on the cheap' they only had a
single scsi controller.  All scsi Hd's were attached to the same
controller as the scsi tape unit.  No raid was used in any form.

One machine is at FBSD 4.3.  It has worked flawlessly for well over 3
years. (it started with linux)  

The other machine is at FBSD 4.8, recently changed from linux.

Now the question is, when I upgraded my second server, I had to change
scsi hardware to an adaptec 29160 as my older buslogix/mylex card was
not supported under freebsd.  I had to go with a wide scsi controller
becuase my tape unit is an external device that requires wide scsi.
After all the hardware was ready, and FBSD was installed, I went to
restore my home directories that were on tape.  My tape unit behaved
poorly and soon I started to get many strange errors from the kernel.  

The last message I caught was
spec_getpages:(#da/0x20000) I/O read failure (error-6) bp 0xc68321bc
vo 0xcd379ec0

after that the system becomes unreadable.

I called the cybernetics people (maker of the tape unit) and their
recommendation was to put the tape unit on a separate scsi controller
from the hard drives.  I didn't want to do this since but did anyway.
It seems to have fixed the problem.  I can now use the tape unit.  The
question is

1) my first FSBD 4.3 system works perfectly find with a single scsi
card and all devices attached to it (though it is a different) model
tape unit.  Why would my second system barf when set up that way.

2) My tape unit and scsi system had been working under linux for years
without a problem.  Does FBSD do scsi so differently?

3) is anyone else running a tape unit and hard drives off the same

4) I thought the whole point of scsi was to allow multiple devices
intelligently communicate on the same buss.  Yet, it seems that is NOT
the case with tape units and hard drives -- the guy at cybernetics
mentioned their access methods can cause problems.

Sorry for the length, but I need a new working paradigm regarding scsi
cause it looks like everything I thought was wrong...


David Bear
phone: 	480-965-8257
fax: 	480-965-9189
College of Public Programs/ASU
Wilson Hall 232
Tempe, AZ 85287-0803
 "Beware the IP portfolio, everyone will be suspect of trespassing"

More information about the freebsd-questions mailing list