Bug #8264
closed
Native scheduler doesn't handle 'du' command failure
Added by Alexey Polushkin almost 7 years ago.
Updated over 6 years ago.
Description
Sometimes Native scheduler writes to console
Process Process-1:
Traceback (most recent call last):
File "/usr/lib64/python3.4/multiprocessing/process.py", line 254, in _bootstrap
self.run()
File "/usr/lib64/python3.4/multiprocessing/process.py", line 93, in run
self._target(*self._args, **self._kwargs)
File "/home/alexey/klever/klever/scheduler/utils/__init__.py", line 288, in disk_controller
size = dir_size("./")
File "/home/alexey/klever/klever/scheduler/utils/__init__.py", line 224, in dir_size
return int(get_output('du -bs {} | cut -f1'.format(dir)))
ValueError: invalid literal for int() with base 10: "du: невозможно получить доступ к './klever-core-work-dir/lkbce/arch/x86/include/asm/linkage.h.lock': Нет такого файла или каталога\n4503272098"
- Status changed from New to Resolved
Resolved in branch 'fix_scheduler_warning'
- Subject changed from Native scheduler doesn't handle 'du' command failuer to Native scheduler doesn't handle 'du' command failure
- Status changed from Resolved to Verified
- Status changed from Verified to Closed
- Status changed from Closed to Open
The issue likely still exists in master since I opened #8349.
Evgeny Novikov wrote:
The issue likely still exists in master since I opened #8349.
Indeed the bug fix isn't correct since another exception occurs when processing the specified one (see #8349 for the traceback).
- Status changed from Open to Resolved
- Priority changed from Urgent to Immediate
Indeed this can break everything unexpectedly, so, this definitely should be in the upcoming release.
- Status changed from Resolved to Closed
I hope that the fixed fix works. So, I merged the branch to master in be75f4f9.
Also available in: Atom
PDF