I noticed this in my Bacula logs:
30-Oct 11:23 syrys-sd JobId 202: End of medium on Volume "BAK101L6" Bytes=2,903,978,804,224 Blocks=1,384,725 at 30-Oct-2025 11:23.
30-Oct 11:23 syrys-sd JobId 202: 3307 Issuing autochanger "unload Volume BAK101L6, Slot 2, Drive 0" command.
30-Oct 11:25 syrys-sd JobId 202: 3995 Bad autochanger "unload Volume BAK101L6, Slot 2, Drive 0": ERR=Child exited with code 1
Results=Traceback (most recent call last):
File "/etc/bacula/scripts/mtx-changer-python/mtx-changer-python.py", line 1095, in <module>
result = unload()
^^^^^^^^
File "/etc/bacula/scripts/mtx-changer-python/mtx-changer-python.py", line 898, in unload
if checkdrive() == 1:
^^^^^^^^^^^^
File "/etc/bacula/scripts/mtx-changer-python/mtx-changer-python.py", line 717, in checkdrive
sg = get_sg_node()
^^^^^^^^^^^^^
File "/etc/bacula/scripts/mtx-changer-python/mtx-changer-python.py", line 603, in get_sg_node
sg_search = re.search('.*' + st + ' .*(/dev/sg\\d+)', result.stdout)
^^
UnboundLocalError: cannot access local variable 'st' where it is not associated with a value
30-Oct 11:25 syrys-dir JobId 202: Recycled volume "BAK103L6"
30-Oct 11:25 syrys-sd JobId 202: 3304 Issuing autochanger "load Volume BAK103L6, Slot 4, Drive 0" command.
From the mtx-changer-python log, it seems like this is related to finding the SG node for the tape drive:
2025-10-30 03:39:04 | ----------[ Starting MTX-Changer-Python v1.32 ]----------
2025-10-30 03:39:04 | Config File: /etc/bacula/scripts/mtx-changer-python/mtx-changer-python.conf
2025-10-30 03:39:04
2025-10-30 11:23:47 | Slot: 2
2025-10-30 11:23:47 | ----------------------
2025-10-30 11:23:49 | ----------[ Starting MTX-Changer-Python v1.32 ]----------
2025-10-30 11:23:49 | Config File: /etc/bacula/scripts/mtx-changer-python/mtx-changer-python.conf
2025-10-30 11:23:49 | Config Section: [DEFAULT]
2025-10-30 11:23:49 | Changer Device: /dev/changer
2025-10-30 11:23:49 | Drive Device: /dev/tape_nst
2025-10-30 11:23:49 | Command: unload
2025-10-30 11:23:49 | Drive Index: 0
2025-10-30 11:23:49 | Slot: 2
2025-10-30 11:23:49 | ----------------------
2025-10-30 11:23:51 - Checking if drive device /dev/tape_nst (drive index: 0) is loaded
2025-10-30 11:23:53 - Drive device /dev/tape_nst (drive index: 0) is loaded with volume (BAK101L6) from slot 2
2025-10-30 11:23:53 - Unloading volume (BAK101L6) from drive device /dev/tape_nst (drive index: 0) to slot 2
2025-10-30 11:25:19 - Successfully unloaded volume (BAK101L6) from drive device /dev/tape_nst (drive index: 0) to slot 2
2025-10-30 11:25:19 - The chk_drive variable is True, calling checkdrive() function
2025-10-30 11:25:19 - Found the following cleaning tapes: [('16', 'CLN001L4')]
2025-10-30 11:25:19 - Determining the tape drive's scsi generic device node required by sg_logs
It doesn't seem to stop the backup from running, but it does seem to interrupt the process of loading a cleaning tape if one is needed.
The cause of the failed assignment seems to be that the "if" statement from 588..598 has no "else" case for detecting a failure to identify the drive's SG device:
|
if '/dev/st' in drive_device or '/dev/nst' in drive_device: |
|
# OK, we caught the simple /dev/st# or /dev/nst# case |
|
# --------------------------------------------------- |
|
st = drive_device |
|
elif '/by-id' in drive_device or '/by-path' in drive_device: |
|
# OK, we caught the /dev/tape/by-id or /dev/tape/by-path case |
|
# ----------------------------------------------------------- |
|
# The ls command outputs a line feed that needs to be stripped |
|
# ------------------------------------------------------------ |
|
st = '/dev/' + re.sub(r'.* -> .*/n*(st\d+).*$', '\\1', result.stdout.rstrip('\n'), re.S) |
|
cmd = lsscsi_bin + ' -g' |
However the root cause is that I've symlinked the drive and use the symlinks in my Bacula configuration:
/dev/changer -> /dev/sg2
/dev/tape_nst -> /dev/nst0
/dev/tape_sg -> /dev/sg1
The if statement ideally needs an else case to detect an unknown type of device node and fail appropriately, but adding support for symlinked device nodes only requires a call to os.path.realpath:
# -----------------------------------------------------------
# TODO: waa - 20240302 - These lines before the if statement
# are not necessary. Probably are here for logging mainly
# -----------------------------------------------------------
real_drive_device = os.path.realpath(drive_device)
log('Resolved drive device to ' + real_drive_device, 20)
cmd = ls_bin + ' -l ' + real_drive_device
log('ls command: ' + cmd, 30)
result = get_shell_result(cmd)
log_cmd_results(result)
chk_cmd_result(result, cmd)
if '/dev/st' in real_drive_device or '/dev/nst' in drive_device:
I noticed this in my Bacula logs:
From the mtx-changer-python log, it seems like this is related to finding the SG node for the tape drive:
It doesn't seem to stop the backup from running, but it does seem to interrupt the process of loading a cleaning tape if one is needed.
The cause of the failed assignment seems to be that the "if" statement from 588..598 has no "else" case for detecting a failure to identify the drive's SG device:
mtx-changer-python/mtx-changer-python.py
Lines 588 to 598 in 7049ff5
However the root cause is that I've symlinked the drive and use the symlinks in my Bacula configuration:
The
ifstatement ideally needs anelsecase to detect an unknown type of device node and fail appropriately, but adding support for symlinked device nodes only requires a call toos.path.realpath: