Richard,
you need to know your linux better :-)
I'll try to give you some hints...
Richard White wrote:
We have Bacula on RH 9 with an internal VXA-1 drive attached to an Adaptec
2940UW and an external RakPak (2 VXA-1 drives) that had been attached to the
external port/channel on the same adapter.
Configured in this way, Bacula would recognize all three drives, but we
encountered a peculiar problem when attempting a full backup on the external
drives. Bacula would write anywhere from 1.5 GB to 24 GB onto a tape, then
fail. Here is the error portion of the log:
.
.
.
24-Jul 09:04 lbackup-sd: GIS_Three.2005-07-24_08:42.29 Error: block.c:552 Write
error at 1:4587 on device /dev/st2. ERR=Input/output error.
24-Jul 09:04 lbackup-sd: GIS_Three.2005-07-24_08:42.29 Error: Error writing
final EOF to tape. This tape may not be readable.
dev.c:1213 ioctl MTWEOF error on /dev/st2. ERR=No such device or address.
ERR=No such device or address.
24-Jul 09:04 lbackup-sd: GIS_Three.2005-07-24_08:42.29 Fatal error: Fatal
device error: ERR=dev.c:639 ioctl MTIOCGET error on /dev/st2. ERR=No such
device or address
24-Jul 08:52 GIS2-fd: GIS_Three.2005-07-24_08:42.29 Fatal error: backup.c:477
Network send error 32768 to SD. ERR=Broken pipe
.
.
.
Now, these same tapes work just fine in the internal drive when I edit
bacula-dir.conf to send this job to that drive.
So we put another 2940UW into the server and attached the RakPak to it. I was concerned
that the drives would be identified differently; I'm a little out of my league here,
though I have a smidgen of understanding. Anyway, although ./btape -c bacula-sd.conf
/dev/st1 and /dev/st2 give a btape prompt, bacula will not mount a tape on either of the
external drives. When I type "mount Left", Bacula instantly informs me:
3905 Device /dev/st1 open but no Bacula volume is mounted.
If this is not a blank tape, try unmounting and remounting the Volume.
It does this without reaching out and touching the drive at all.
When I run ./btape -c bacula-sd.conf /dev/st1, it opens the device OK. Then I type
"test" and get this:
btape: btape.c:766 Bad status from rewind. ERR=dev.c:406 Rewind error on
/dev/st1. Err=No medium found.
Bulletin: While writing this I have done a number of tests and I now have some
more details.
When I start Bacula and type status sd, it shows that Week_1B is mounted on Internal
(/dev/st0) and Tuesday_A is mounted on Right (/dev/st2). This is interesting because
Tuesday_A is in the internal drive and Week_1B is in the left drive. That implies that
Bacula now thinks that /dev/st1 is the Left drive, meaning that the drives have
"rotated" one step clockwise.
Moreover, if I unmount all drives, stop Bacula, then restart and run status sd,
the tapes are both mounted again as before. As before, if I unmount and then
mount, I get the errors mentioned above.
So I stopped Bacula, edited bacula-dir.conf slightly to send a full backup to the "Internal" drive
(formerly sent to "Left"), then restarted and am currently running a full backup on the
"Internal" (Left RakPak) tape drive. I will know in a couple of hours if it finishes successfully.
Even if it does, there remains the problem of unmounting and remounting. I have little doubt that if I edit
bacula-sd.conf all will be corrected, but I have no idea what to change.
Summary: You got your device names mixed up. I think.
Background:
Tape drives are numbered in the order they are noted by the kernel.
This order depends on a number of variables:
Basically, the SCSI subsystem scans all devices at startup (you can add
devics later, and so on, but we leave this for later ;-)
Usually, it scans the SCSI IDs from one upwards, so a tape device with
ID 2 might become st0 and one with ID 4 st1.
With multiple SCSI HBAs or busses per HBA things become more
interesting, because the scans happen per SCSI bus. So, to know in which
order the devices are found, you need to know the order the SCSI drivers
are loaded, the order in which they find their adapters, tho order in
which they access the busses. devfs and the related daemons like hal and
subfs can make this even more interesting, by the way, but up till now
the keep their fingers off SCSI and tape drives - as far as I know.
All this can be somewhat difficult, so I suggest you don't try that...
instead, use the practical part of this tutorial :-)
stop bacula.
(Now, I assume you don't know the serial numbers of your drives... then
you can find their order simply by doing a 'cat /proc/scsi/scsi' ...)
load a tape in only one of your drives. Wait till the drive has loaded
the tape and all lights are green.
Do a mt -f /dev/nst0 status. Then, the same with st1 and st2.
Only one of the commands should result in output describing a loaded
tape drive.
Put the tape in one of the remaining drives - you know what to do then,
I guess.
After that, you have the linux device names for your three drives.
Then, modify the bacula-sd.conf to assign the right device names to your
defined storage devices.
Hope this helps,
Arno
Richard
-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
from IBM. Find simple to follow Roadmaps, straightforward articles,
informative Webcasts and more! Get everything you need to get up to
speed, fast. http://ads.osdn.com/?ad_idt77&alloc_id492&op=click
_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users
--
IT-Service Lehmann [EMAIL PROTECTED]
Arno Lehmann http://www.its-lehmann.de
-------------------------------------------------------
SF.Net email is sponsored by: Discover Easy Linux Migration Strategies
from IBM. Find simple to follow Roadmaps, straightforward articles,
informative Webcasts and more! Get everything you need to get up to
speed, fast. http://ads.osdn.com/?ad_id=7477&alloc_id=16492&op=click
_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users
