Re: [Bacula-users] Re: [bacula 0000090]: Tape drive snafu confused bacula job status reporting

[Martin S. <ma...@xa...>]

>>>>> On Tue, 24 Aug 2004 17:14:10 +0200, Kern Sibbald <ke...@si...> said:

  Kern> On Tue, 2004-08-24 at 16:51, Martin Simmons wrote:
  Kern> ...
  >> 
  >> Thanks, that looks good -- I'll run with that patch.  As you guessed, I don't
  >> want to crash the tape drive again by trying to repeat it!  The underlying
  >> problem was fastfsf being unreliable so I have switched that off in the config
  >> now.

  Kern> Huh? I missed that. What was unreliable about fastfsf? -- I can believe
  Kern> it, but would like to understand what you mean.

Regardless of the eot model (1 or 2), I found that leaving the device open for
more than 30 minutes (exactly) would make fsf 32767 (i.e. fastfsf) fail.  My
guess is that tape drive goes into some kind of sleep mode and doesn't quite
wake up correctly.  It would usually fail at the end of the last tape file,
but it took around an hour to fail on a tape that I had used for the rawfill
test (in the past), so I suspect it was scanning the whole tape.

E.g. with a rewound tape containing a successful btape test run

./btape /dev/nsa0
Tape block granularity is 1024 bytes.
btape: butil.c:168 Using device: "/dev/nsa0" for writing.
btape: btape.c:310 open_dev /dev/nsa0 OK
<<<<<<<<<<<<< I waited 30 minutes here
*fsf 32767
btape: btape.c:1400 Bad status from fsf. ERR=dev.c:808 ioctl MTFSF error on /dev/nsa0. ERR=Input/output error.

Closing the device and rewinding usually fixed it.

The btape test command worked fine of course because it runs in a few minutes.


  Kern>                                                  It seemed to me that
  Kern> your tape drive ran into the physical end of the tape while writing or
  Kern> got a *hard* I/O error.

That last time it nearly ran into end of tape *drive* :-)  I had to power cycle
the machine to get it to be visible on the SCSI bus again.

In general, it wasn't a real I/O error though, because I could always fix it
by using the unmount and mount commands in the console.

I am now back to using the "normal" FreeBSD configuration, which seems to be
working:

  AutomaticMount = yes;
  AlwaysOpen = yes
  Offline On Unmount = no
  Hardware End of Medium = no
  BSF at EOM = yes
  Backward Space Record = no
  Backward Space File = no
  Fast Forward Space File = no
  TWO EOF = yes

__Martin