Project

General

Profile

Actions

Bug #11994

closed

[arv-mount] Do not crash if /sys/fs/fuse/connections is empty

Added by Tom Clegg over 7 years ago. Updated about 7 years ago.

Status:
Resolved
Priority:
Normal
Assigned To:
Category:
FUSE
Target version:
Story points:
-

Description

In debian 9 (stretch), due to a systemd-blames-kernel bug, /sys/fs/fuse/connections/ might be empty.

When this occurs, instead of crashing, arv-mount --unmount and friends should print a warning and try the next best thing.

The current behavior is a crash with a stack trace culminating in:

IOError: [Errno 2] No such file or directory: '/sys/fs/fuse/connections/45/abort'

Subtasks 1 (0 open1 closed)

Task #12563: Review 11994-no-sysfs-fuseResolvedPeter Amstutz07/19/2017Actions

Related issues 2 (0 open2 closed)

Related to Arvados - Bug #12306: [arv-mount] --unmount should work on an unresponsive mountResolvedTom Clegg09/22/2017Actions
Related to Arvados - Bug #12538: crunch-run failing to terminate after completeResolvedPeter Amstutz11/06/2017Actions
Actions #1

Updated by Peter Amstutz about 7 years ago

Instead of crashing, arv-mount --unmount and friends should print a warning and try the next best thing.

I'm not sure what the next best thing is, probably kill -9 arv-mount and then try fusermount -u -z again?

Actions #2

Updated by Tom Clegg about 7 years ago

  • Status changed from New to In Progress
  • Assigned To set to Tom Clegg
  • Target version changed from Arvados Future Sprints to 2017-11-08 Sprint

I'm not sure what the next best thing is, probably kill -9 arv-mount and then try fusermount -u -z again?

Without the sysfs dir, we can't look up the pid, so I don't think we can send a signal with kill either. So just fall through to "fusermount -u -z" and try again, yes.

Turns out the "ignore ENOENT" code has been there since the beginning of unmount.py, but it didn't work because it was looking for OSError instead of IOError.

11994-no-sysfs-fuse @ 4a8561a37ef5878aefc8c5d8c47898a3879a0fda

Actions #3

Updated by Peter Amstutz about 7 years ago

Tom Clegg wrote:

I'm not sure what the next best thing is, probably kill -9 arv-mount and then try fusermount -u -z again?

Without the sysfs dir, we can't look up the pid, so I don't think we can send a signal with kill either. So just fall through to "fusermount -u -z" and try again, yes.

So you don't want to walk the process table and look for an arv-mount with the target path in its arguments? (yes, sort of joking)

Turns out the "ignore ENOENT" code has been there since the beginning of unmount.py, but it didn't work because it was looking for OSError instead of IOError.

11994-no-sysfs-fuse @ 4a8561a37ef5878aefc8c5d8c47898a3879a0fda

LGTM.

Actions #4

Updated by Anonymous about 7 years ago

  • Status changed from In Progress to Resolved

Applied in changeset arvados|commit:d493d06b4b93b87c10c59954e47a18d85db13150.

Actions

Also available in: Atom PDF