Red Hat Bugzilla – Bug 469116
virt-manager silently exits when guest installation completes or is killed
Last modified: 2010-03-16 13:15:58 EDT
Description of problem:
Whenever a guest that is being installed terminates, either because installation has completed and it's rebooting, or the administrator has forcibly killed the guest, virt-manager exits, taking out all open guest consoles.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Install Fedora 10 Snap 2 and update to latest. (Installing F10-Snap3 would probably suffice.)
2. Start an installation with virt-manager.
3. Kill the installation with Force Shutdown.
virt-manager disappears, along with any windows it had opened, including unrelated guests. No error message is printed on the console virt-manager was launched from.
Once the guest has been installed, killing it with Force Shutdown has no adverse effects.
virt-manager survives. The VM window should probably survive too.
Only tested invoking virt-manager over ssh -X from up-to-date Fedora 8 machines, but no X messages appear on the console other than the following, which appears when starting virt-manager and seems to have no adverse effects on this or any other X app:
Xlib: extension "RANDR" missing on display "localhost:10.0".
Can you try to reproduce running virt-manager with --no-fork from the command line and see if something is printed when the app closes? I'll try to reproduce as well.
I tried running with --no-fork, this time ssh-ing in from a box that's running F9 plus whatever rawhide packages that rawhide virt tools pull in. When I did the force poweroff, some message about a failed TCP connection flashed momentarily in the GUI window, and then it shows "guest not running", as would be expected, and nothing died that shouldn't have. Nothing was printed on the console when I killed the guest.
I'll try again sshing in from an F8 box tonight. I'm guessing this isn't a problem running locally, otherwise everyone running virt-manager on rawhide would be seeing it. This could just be an F8 bug, but it could also be that virt-manager or dbus-x11 is assuming too much about the X server.
Interesting. Running with --no-fork when SSHd in from F8 (where the problem happens) causes the same behavior as my F9 test. I never actually tested F9 without --no-fork. It appears that --no-fork makes the problem go away. Fun.
Hmm. There was a problem similar to this where the app wouldn't start: it was related to dbus and X. I'll try and reproduce and see if I can gather anything.
Any chance you are running F10 or rawhide? There was a fix in this area that I think may solve the issue. Wondering if you could give it a test?
I'm pretty certain this was solved with fixes applied to virt-manager 0.6.0 in f9 and f10, so closing as CURRENTRELEASE. Please reopen if you can still reproduce, and include
local and remote host distros
result of trying with and without --no-fork