Opened 22 months ago
Closed 22 months ago
#310 closed defect (fixed)
qubes-manager silently crashes every once in a while
| Reported by: | joanna | Owned by: | rafal |
|---|---|---|---|
| Priority: | critical | Milestone: | Release 1 Beta 2 |
| Component: | core | Keywords: | |
| Cc: |
Description
The actual crash error msg is very strange (obtained by running it from commandline):
Fatal Python error: deallocating None Aborted
This applies to v1.1.2, but the changes since v1.1.1 were trivial and are likely not the cause. Instead, probably some changes to core got us into some race condition. Perhaps it causes some refcounts to be incorrectly managed... This really is strange.
Change History (4)
comment:1 Changed 22 months ago by joanna
comment:2 Changed 22 months ago by rafal
- Owner changed from joanna to rafal
- Status changed from new to assigned
Is there any correlation between user actions and manager crash ? Or does it crash without user input ? Any correlation to other events (vm start/stop) ?
comment:3 Changed 22 months ago by joanna
I was unable to find an exact event that forces this crash. What I observed:
1) it happened a few times in a row when I was making a backup (qvm-backup)
2) it happened a few times when I was starting/stopping a VM
3) it doesn't seem to be related to querying the qubes DB (e.g. I run qvm-ls in a loop and nothing happened).
comment:4 Changed 22 months ago by rafal
- Resolution set to fixed
- Status changed from assigned to closed
Most probably fixed in
http://git.qubes-os.org/?p=rafal/xen.git;a=commit;h=35c8e3baeb46f64c7c58e8f952b5ff2be3a5fe37
prebeta2 branch.
Marek, you may wish to rectify this upstream ?
I'm running core 1.6.11 BTW. No problems observed when running core 1.6.10:
http://git.qubes-os.org/?p=joanna/core.git;a=commit;h=6715759e32b8ef638c12ac53a601b0348276a972