Backups to tape produce input/output errors on Debian kernel 6.1.0-37-amd64

[jens.gr...@gmail.com]

Hello folks,

since 2025-05-26 I'm experiencing errors when writing backups from disk to tape. Here's the first of several error messages in the log:

26-Mai 23:16 bareos-sd JobId 61405: Error: stored/block.cc:750 Write error at 30:8296 on device "IBM-Drive-0" (/dev/nst0). ERR=Eingabe-/Ausgabefehler.
26-Mai 23:16 bareos-sd JobId 61405: Error: stored/block.cc:768 Write error on fd=5 at file:blk 30:8296 on device "IBM-Drive-0" (/dev/nst0). ERR=Eingabe-/Ausgabefehler.
26-Mai 23:17 bareos-sd JobId 61405: Error: Re-read of last block OK, but block numbers differ. Read block=8294 Want block=8295.
26-Mai 23:17 bareos-sd JobId 61405: End of medium on Volume "WOC4-1L78" Bytes=1,155,880,771,584 Blocks=275,597 at 26-Mai-2025 23:17.
26-Mai 23:17 bareos-sd JobId 61405: 3307 Issuing autochanger "unload slot 1, drive 0" command.
26-Mai 23:19 bareos-dir JobId 61405: Using Volume "WOC1-4L78" from 'Scratch' pool.

I'm using bareos 24.0.4~pre0.1014be830-74 (community repo) on an up-to-date Debian 12.11. The one thing that has happened when the first error occurred was a kernel upgrade on the computer from version 6.1.0-35-amd64 to 6.1.0-37-amd64. I rebooted the computer after the upgrade.

What I've tried so far:

• I thought this was a problem with an old tape so in the meantime I tried with 5 different tapes which produce the same error.
• I ran some drive check in the storage loader's Web-UI but no problems or errors are shown.
• I've booted into the old kernel and I'm running the backup again. The backup is currently running for more than an hour without errors.

I would now wait for the backup to finish and then try again with the newer kernel to see if the problem persists.

Has anyone experienced similar problems after the kernel upgrade?

Greetings, Jens

[jens.gr...@gmail.com]

The backup finished with the old kernel without any errors.

After that I rebooted into the new kernel and started another (smaller) backup job which also ran without errors.

So maybe it was just a weird coincidence. I will let some more backups run with the new kernel and report if the errors occur.

Greetings

Jens

[Andreas Rogge]

Hi Jens,

Am 03.06.25 um 10:11 schrieb jens.gr...@gmail.com:

> The backup finished with the old kernel without any errors.
>
> After that I rebooted into the new kernel and started another (smaller)
> backup job which also ran without errors.
>
> So maybe it was just a weird coincidence. I will let some more backups
> run with the new kernel and report if the errors occur.

It usually makes sense to look at the kernel-log (i.e. dmesg) after
things like that happen.
While the tape driver's API is pretty terse and doesn't usually report
anything beyond "IO Error" the kernel log is usually a lot more chatty.

Best Regards,
Andreas
--
Andreas Rogge andrea...@bareos.com
Bareos GmbH & Co. KG Phone: +49 221-630693-86
http://www.bareos.com

Sitz der Gesellschaft: Köln | Amtsgericht Köln: HRA 29646
Komplementär: Bareos Verwaltungs-GmbH
Geschäftsführer: Stephan Dühr, Jörg Steffens, Philipp Storz

[jens.gr...@gmail.com]

Hi Andreas,

thank you for the tip. I've looked into the kernel.log and found these entries:

2025-05-26T23:16:43.264361+02:00 gandalf kernel: [49612.006103] mpt3sas_cm0: log_info(0x31120439): originator(PL), code(0x12), sub_code(0x0439)
2025-05-26T23:16:43.264389+02:00 gandalf kernel: [49612.006153] st 1:0:0:0: [st0] Error b0000 (driver bt 0, host bt 0xb).
…

2025-05-27T23:09:14.578491+02:00 gandalf kernel: [135560.892293] mpt3sas_cm0: log_info(0x31120439): originator(PL), code(0x12), sub_code(0x0439)

2025-05-27T23:09:14.578508+02:00 gandalf kernel: [135560.892324] st 1:0:0:0: [st0] Error b0000 (driver bt 0, host bt 0xb).

I'm not sure what that exactly means, but some research and ChatGPT make me think that this may indicate a medium error, a hardware issue, or a cable/SAS connection problem.

I've tested different tapes with the same error so i would rule out the medium error. The tape drive was replaced in November 2024 so I would rule that out too. Maybe a cable/SAS connection problem which has been magically resolved by the reboot.

As I wrote I will have an eye on the next few backups hoping I can close this issue.

Greetings, Jens

[jens.gr...@gmail.com]

Hello again,

the latest backup have been running smoothly without any errors so the issue can be closed.

Greetings, Jens