arch linux running inside virtualbox on a win10 host accessing borg repo via vboxsf (which is known-problematic). borg repo resides on host's ntfs filesystem.
this issue is not about vboxsf being unreliable, but rather about checking whether we can make borg check more robust to not crash here:
~ borg check --debug -p /mnt/z/borg/borg.repo
using builtin fallback logging configuration
35 self tests completed in 0.06 seconds
Starting repository check
Upgrading from v1 hints.31692
Rebuilding sparse info for segment 28722
Failed to read committed index ([Errno 2] No such file or directory: '/mnt/z/borg/borg.repo/data/2/28722')
Segment transaction is 31693
Determined transaction is 31693
Found 26881 segments
Local Exceptionts 93.8%
Traceback (most recent call last):
File "/home/doc/.local/lib/python3.6/site-packages/borg/archiver.py", line 4175, in main
exit_code = archiver.run(args)
File "/home/doc/.local/lib/python3.6/site-packages/borg/archiver.py", line 4107, in run
return set_ec(func(args))
File "/home/doc/.local/lib/python3.6/site-packages/borg/archiver.py", line 150, in wrapper
return method(self, args, repository=repository, **kwargs)
File "/home/doc/.local/lib/python3.6/site-packages/borg/archiver.py", line 304, in do_check
if not repository.check(repair=args.repair, save_space=args.save_space):
File "/home/doc/.local/lib/python3.6/site-packages/borg/repository.py", line 904, in check
objects = list(self.io.iter_objects(segment))
File "/home/doc/.local/lib/python3.6/site-packages/borg/repository.py", line 1317, in iter_objects
read_data=read_data)
File "/home/doc/.local/lib/python3.6/site-packages/borg/repository.py", line 1398, in _read
data = fd.read(length)
OSError: [Errno 71] Protocol error
Platform: Linux staskij 4.14.10-1-ARCH #1 SMP PREEMPT Fri Dec 29 20:17:35 UTC 2017 x86_64
Linux: arch
Borg: 1.1.4 Python: CPython 3.6.4
PID: 785 CWD: /home/doc
sys.argv: ['/home/doc/.local/bin/borg', 'check', '--debug', '-p', '/mnt/z/borg/borg.repo']
SSH_ORIGINAL_COMMAND: None
from irc user mlazze:
arch linux running inside virtualbox on a win10 host accessing borg repo via vboxsf (which is known-problematic). borg repo resides on host's ntfs filesystem.
this issue is not about vboxsf being unreliable, but rather about checking whether we can make borg check more robust to not crash here: