Ticket #1500 (closed defect: fixed)

Opened 7 years ago

Last modified 7 years ago

gvfs prevents Precise chroots from ending cleanly

Reported by: jdreed Owned by:
Priority: normal Milestone: Summer 2014
Component: -- Keywords:
Cc: Fixed in version:
Upstream bug:

Description

For a while, we've been hearing sporadic reports that workstations are rebooting after every single user. I finally reproduced this on my VM. The culprit is gvfs' fuse daemon. Even if unmounted cleanly inside the chroot, schroot will insist it cannot umount AFS due to EBUSY. Indeed, setting export GVFS_DISABLE_FUSE=1 in snapshot-run makes everything work again.

Another solution is passing -l to the umount command for the bind-mount of AFS. In theory, this should be fine, and is never the wrong answer, but I'd like someone more familiar with AFS than me to chime in if they think that is likely to have unexpected failure modes. If no objections, I will commit an schroot setup.d script that does that.

(Yes, Precise is going away, but a) cluster machines need to be up and running to get upgraded to Trusty; b) we should in fact fix this anyway.)

Change History

comment:1 Changed 7 years ago by jdreed

  • Status changed from new to committed

committed  929d139164f72496dfd3dddf6938873373241a83 (Add setup.d to umount -l /afs) to master

comment:2 Changed 7 years ago by jdreed

  • Status changed from committed to development

comment:3 Changed 7 years ago by jdreed

  • Status changed from development to closed
  • Resolution set to fixed

I'd still like to understand why gvfs is being dumb, but meh.

Note: See TracTickets for help on using tickets.