Tape backup error

[jcz]

Hello.

We have some error of tape backup (Tape backup - Content - New backup). No changer, HP-Ultrium-6250 drive only.

PBS GUI:

2021-06-30T10:25:15+02:00: update media online status
2021-06-30T10:25:15+02:00: media set uuid: ...
2021-06-30T10:25:15+02:00: found 2 groups
2021-06-30T10:25:15+02:00: backup snapshot vm/101/2021-06-22T22:30:01Z
2021-06-30T10:25:15+02:00: allocated new writable media 'weekly-XX'
2021-06-30T10:26:09+02:00: found media label weekly-XX (...)
2021-06-30T10:26:09+02:00: TapeAlertFlags: READ_FAILURE
2021-06-30T10:26:09+02:00: TASK ERROR: aborting due to critical tape alert flags: READ_FAILURE

Syslog:

Jun 30 10:26:09 pbs proxmox-backup-proxy[858]: could not send chunk to reader thread: sending on a closed channel

After that:

root@pbs:~# mt -f /dev/nst0 status
/dev/nst0: Input/output error

Tape is "damaged", but can be formatted and labeled (after reboot). Also i can write and read to/from by dd.

Any idea?

Thx.

J.

 

[dietmar]

Hello.

We have some error of tape backup (Tape backup - Content - New backup). No changer, HP-Ultrium-6250 drive only.

PBS GUI:

2021-06-30T10:25:15+02:00: update media online status
2021-06-30T10:25:15+02:00: media set uuid: ...
2021-06-30T10:25:15+02:00: found 2 groups
2021-06-30T10:25:15+02:00: backup snapshot vm/101/2021-06-22T22:30:01Z
2021-06-30T10:25:15+02:00: allocated new writable media 'weekly-XX'
2021-06-30T10:26:09+02:00: found media label weekly-XX (...)
2021-06-30T10:26:09+02:00: TapeAlertFlags: READ_FAILURE
2021-06-30T10:26:09+02:00: TASK ERROR: aborting due to critical tape alert flags: READ_FAILURE

This indicates that the cartridge is damaged (stored in cartridge memory). Have you already tried with another cartridge?

Also, please use 'pmt' instead of 'mt' command.

 

[jcz]

Not yet, will try. Thanks.

 

[jcz]

So... same with other tape

root@pbs:~# pmt status
using device /dev/tape/by-id/scsi-HUJ6XXXCAV-sg
Error: SCSI inquiry failed - scsi command failed: transport error

 

[dcsapak]

ok normally that read_failure flag gets only set if either the tape or the drive produces read errors (meaning one of them could be damaged)

Also i can write and read to/from by dd.

where to you read write to ?

can you read the exact data back when you write it ?
e.g. try with random data and check if you get the same data back

 

[jcz]

Tested with:

dd if=random.bin of=/dev/nst0 bs=512k
dd if=/dev/nst0 bs=512k of=random.XX

Several times with hash checks. All tests were fine.

 

[dcsapak]

mhmm how much data did you write on the tape ? (maybe it was to little and was cached by the drive ?)

 

[jcz]

About 5x 1GB. Now i tried 5GB in one file with dd and had almost same error. Maybe drive is really bad