Subject: Re: scsi tape 'Not Ready to Ready Transition'
To: Andreas Wrede <email@example.com>
From: Jan-Hinrich Fessel <firstname.lastname@example.org>
Date: 10/21/1997 08:37:30
Content-Type: text/plain; charset=us-ascii
In message <email@example.com>you write:
> I get the following scsi check condition on every first access to the tape
> after a tape change:
> st0(ncr0:4:0): Check Condition on opcode 0x0
> SENSE KEY: Unit Attention
> EOM Detected
> ASC/ASCQ: Not Ready To Ready Transition (Medium May Have Changed)
> In other words, the sequence
> $ mt offl
> <insert new tape and wait for drive to become ready>
> $ mt status
> mt: /dev/nrst0: Input/output error
> st0 at scsibus0 targ 4 lun 0: <EXABYTE, EXB-8500SMBANXH1, 0458> SCSI2 1/sequential removable
> st0: st0(ncr0:4:0): 4.0 MB/s (250 ns, offset 8)
It is the drive, that is set up to report medium changes. This is correct, as
I belive, but the driver in NetBSD does not handle this.
I don't have a fix for this problem handy, but I suggest you either
a) live with it and do mt status after changing tapes before any other action
b) disable the reporting in the drive (which should be possible, at least the
EXB 8200 had a switch for that) or
c) get a fix for the driver ready or
d) wait til I send-pr a fix or
e) do whatever you like ;-)
c) and d) are not that easy as they look at first shot, because the medium
change should be visible to software that uses it actively and transparent for
software knows not how to handle it. I am not sure if we need another minor
device node set or if we can get away with ioctl/fcntl style setup or even
with sysctl and a per drive default.
And I have not looked at how the "others" handle that, so any input on this
would be nice.
Tragbar ist, was nicht herunterfaellt.
-----BEGIN PGP MESSAGE-----
-----END PGP MESSAGE-----