On Tue, Apr 11, 2006 at 08:11:02AM -0700, Jason Martin wrote:
> On Tue, Apr 11, 2006 at 04:17:50PM +0200, Kern Sibbald wrote:
> > On Tuesday 11 April 2006 15:46, Jason Martin wrote:
> > >
> > > Now, it looks to me like mtx is opening the changer device
> > > /dev/sg4 (which is the correct device), reading, and getting a
> > > series of insert statements for the File table for MySQL instead
> > > of the device status. 
FYI, I've been able to work around this problem by adding
$MTX -f $ctl status >/dev/null 2>&1
to the mtx-changer script prior to attempting the operation
requested. Doing so seems to flush out the errant data in the
buffers.

-Jason Martin


> > Or possibly it is getting data that should be written to a tape. You aren't 
> > by 
> > any chance using  "Archive device = /dev/sg4" somewhere in one of your 
> > Device 
> > resources are you?
> Thats an idea, but I'm not seeing that in the config.
> Here is my storage config:
> Device {
>   Name = Drive0
>   Media Type = ULTRIUM-TD3
>   Archive Device = /dev/nst0
>   Drive Index = 0
>   AutomaticMount = yes;               # when device opened, read it
>   AlwaysOpen = yes;
>   RemovableMedia = yes;
>   RandomAccess = no;
>   Changer Command = "/etc/bacula/mtx-changer %c %o %S %a %d"
>   Changer Device = /dev/sg2
>   AutoChanger = yes
>   Spool Directory = /data/backups
>   Alert Command = "sh -c 'tapeinfo -f %c |grep TapeAlert|cat'"
> }
> 
> Device {
>   Name = Drive1
>   Media Type = ULTRIUM-TD3
>   Archive Device = /dev/nst1
>   Drive Index = 0
>   AutomaticMount = yes;               # when device opened, read it
>   AlwaysOpen = yes;
>   RemovableMedia = yes;
>   RandomAccess = no;
>   Changer Command = "/bin/sh -c 'strace -s 999 -f -o /tmp/mtx.log.$$ 
> /etc/bacula/mtx-changer %c %o %S %a %d'"
>   Changer Device = /dev/sg4
>   AutoChanger = yes
>   Spool Directory = /data/backups
>   Alert Command = "sh -c 'tapeinfo -f %c |grep TapeAlert|cat'"
> }
> 
> Bacula is configured to run only one job at a time, and both
> drives are functional. Verifies succeed (after getting through
> the tape change issues) so I am confident the data is making it
> to the tape.
> 
> > > Running 'mount' at the bconsole prompt is 
> > > enough to get it to retry the loaded / unload / load cycle for
> > > the next tape and continue.
> 
> > > This is on a ADIC Scalar24 w/LTO-3 drives, using a fiberchannel 
> > > connection.
> > > mtx version 1.2.17rel
> > > Red Hat Enterprise Linux ES release 4 (Nahant Update 2)
> > > bacula-mysql-1.36.3-1
> > > Linux xxx 2.6.9-22.ELsmp #1 SMP Mon Sep 19 18:32:14 EDT 2005 i686 athlon
> > > i386 GNU/Linux QLogic ISP23xx FC-SCSI Host Bus Adapter driver
> > > mysql  Ver 14.7 Distrib 4.1.12, for redhat-linux-gnu (i386) using readline
> > > 4.3
> 
> Thank you,
> -Jason Martin

-- 
He's dead, Jim. Get his ears! - Spock
This message is PGP/MIME signed.

Attachment: pgpMZKgxPK7ov.pgp
Description: PGP signature

Reply via email to