"Nicholas Waltham" <nwaltham@yahoo.com> writes:
| makes a mess of the estimate of compression, and tries to
| put on a tape more than will actually fit. Amanda apparently fails
| gracefully leaving the untaped portion on the disk. However I have
| noticed that, even after changing the tape amanda always says the tape
| is full, and I cannot do any more amflushes or amdumps until I reboot
| the computer. Is this likely to be an amanda problem or a linux kernel
| problem? Has anyone else seen this? I am using a SCSI tape drive - the
>From those two choices I'd vote for Linux kernel problem. I'm
using afbackup and faced exactly the same problem (although I'm not
sure if it goes so that /any/ error situation on tape drive puts it
to a state that requires reboot). I can also reproduce it easily
with just dd and mt. I.e. if an error (at least an attempt to
write more than actually fits qualifies) happens, I can't do
_anything_ with that device; mt and all other software immediately
says `Input/output error'. Also, the tape drive (Seagate Travan,
SCSI model) goes to a state where even removing the old tape and
inserting a new one doesn't invoke that drive's normal
*squeek*-*squeek*-go-forward-go- backward-and-blink-leds procedure.
It simply doesn't do anything. Soft boot doesn't help, the machine
has to be reseted using its reset-switch or possibly even power
turned off and back on (some time has passed since I fiddled with
it so I don't remember for sure).
I concluded that the tape drive is broken and sent it back for
repairing, but things are looking different now that I saw your
mail :( I can't remember with what kernel I first saw this problem
(it was at the end of the summer), but at least it has occured with
2.2.13 and 2.2.14pre series. My Travan is attached to Adaptec
2940UW (I don't know why /proc/pci says Adaptec AIC-7881U (rev 0)
-- perhaps that's the actual chip or something...) in addition to
one hard disk. The hard disk works just fine. The machine is
somewhat old PII 266. I know next to nothing about SCSI so I can't
even start making guesses where the problem might be.
Anyone?
-- Hannu- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Sat Jan 01 2000 - 23:11:58 EST