GNOME Bugzilla – Bug 752955
gnome-boxes occasionally fails to start live CD
Last modified: 2018-01-11 10:25:35 UTC
Not sure if it's Boxes' bug but from time to time I see failing test in automation. It's caused by turning off live cd just after creation. This leads to machine removal and test failure. I am on latest jhbuild version (3.17.4.2-01b1c) but I see this for ages (since 3.14 version). I am in nested KVM environment so the bug may be there or in libvirt. Really, not sure if Boxes are the root cause. There is an error in libvirt logs: Jul 28 10:07:02 localhost.localdomain libvirtd[20101]: internal error: End of file from monitor logs from Boxes: [test@localhost tests]$ grep 55 /tmp/boxes.txt (gnome-boxes:8324): Boxes-DEBUG: vm-creator.vala:277: Creating volume 'boxes-unknown-55'.. (gnome-boxes:8324): Boxes-DEBUG: vm-creator.vala:279: Created volume 'boxes-unknown-55'. (gnome-boxes:8324): Boxes-DEBUG: libvirt-machine.vala:153: new libvirt machine: boxes-unknown-55 (gnome-boxes:8324): Boxes-DEBUG: machine.vala:100: State of 'Core-5 55' changed to BOXES_MACHINE_MACHINE_STATE_STOPPED (gnome-boxes:8324): Boxes-DEBUG: libvirt-broker.vala:134: New domain 'boxes-unknown-55' (gnome-boxes:8324): Boxes-DEBUG: machine.vala:100: State of 'Core-5 55' changed to BOXES_MACHINE_MACHINE_STATE_RUNNING (gnome-boxes:8324): Boxes-DEBUG: libvirt-machine.vala:345: enable statistics for Core-5 55 (gnome-boxes:8324): Boxes-DEBUG: machine.vala:313: Scheduling autosave for 'Core-5 55' (gnome-boxes:8324): Boxes-DEBUG: machine.vala:326: Unscheduling autosave for 'Core-5 55' (gnome-boxes:8324): Boxes-DEBUG: machine.vala:313: Scheduling autosave for 'Core-5 55' [test@localhost tests]$ grep 60 /tmp/boxes.txt (gnome-boxes:8324): Boxes-DEBUG: vm-creator.vala:277: Creating volume 'boxes-unknown-60'.. (gnome-boxes:8324): Boxes-DEBUG: vm-creator.vala:279: Created volume 'boxes-unknown-60'. (gnome-boxes:8324): Boxes-DEBUG: libvirt-machine.vala:153: new libvirt machine: boxes-unknown-60 (gnome-boxes:8324): Boxes-DEBUG: machine.vala:100: State of 'Core-5 60' changed to BOXES_MACHINE_MACHINE_STATE_STOPPED (gnome-boxes:8324): Boxes-DEBUG: libvirt-broker.vala:134: New domain 'boxes-unknown-60' (gnome-boxes:8324): Boxes-DEBUG: machine.vala:100: State of 'Core-5 60' changed to BOXES_MACHINE_MACHINE_STATE_RUNNING (gnome-boxes:8324): Boxes-DEBUG: libvirt-machine.vala:345: enable statistics for Core-5 60 (gnome-boxes:8324): Boxes-DEBUG: machine.vala:313: Scheduling autosave for 'Core-5 60' (gnome-boxes:8324): Boxes-DEBUG: machine.vala:326: Unscheduling autosave for 'Core-5 60' (gnome-boxes:8324): Boxes-DEBUG: machine.vala:313: Scheduling autosave for 'Core-5 60' (gnome-boxes:8324): Boxes-DEBUG: machine.vala:100: State of 'Core-5 60' changed to BOXES_MACHINE_MACHINE_STATE_STOPPED (gnome-boxes:8324): Boxes-DEBUG: libvirt-machine.vala:364: disable statistics for Core-5 60 full log attached in boxes.txt
Created attachment 308292 [details] full Boxes' log
Created attachment 308293 [details] full libvirtd's log
the issue happened around Jul 28 11:35:24 localhost.localdomain libvirtd[9313]: internal error: End of file from monitor
* Can you manually reproduce it? * Please try to reproduce on baremetal. * What exactly happens? You create a live box and it fails to start?
(In reply to Zeeshan Ali (Khattak) from comment #4) > * Can you manually reproduce it? > yes > * Please try to reproduce on baremetal. > will do > * What exactly happens? You create a live box and it fails to start? Box is shutted down so it disappears. Can this be a liveCD issue? I've tried a bunch of them (version 5/6 of tiny core Linux)
(In reply to vladimir benes from comment #5) > (In reply to Zeeshan Ali (Khattak) from comment #4) > > * Can you manually reproduce it? > > > yes > > > * Please try to reproduce on baremetal. > > > will do > > > * What exactly happens? You create a live box and it fails to start? > Box is shutted down so it disappears. Can this be a liveCD issue? I've tried > a bunch of them (version 5/6 of tiny core Linux) From all the input so far, I'd say it's like a qemu crash. Might be related to nested kvm so that's why I asked if you can produce on baremetal.
(In reply to Zeeshan Ali (Khattak) from comment #6) > (In reply to vladimir benes from comment #5) > > (In reply to Zeeshan Ali (Khattak) from comment #4) > > > * Can you manually reproduce it? > > > > > yes > > > > > * Please try to reproduce on baremetal. > > > > > will do > > > > > * What exactly happens? You create a live box and it fails to start? > > Box is shutted down so it disappears. Can this be a liveCD issue? I've tried > > a bunch of them (version 5/6 of tiny core Linux) > > From all the input so far, I'd say it's like a qemu crash. Might be related > to nested kvm so that's why I asked if you can produce on baremetal. Actually I cannot reproduce on bare. I've hit several more crashes but not this one. So nested KVM, very likely. not sure if related to fedora so filed there: https://bugzilla.redhat.com/show_bug.cgi?id=1247941 https://bugzilla.redhat.com/show_bug.cgi?id=1247945
(In reply to vladimir benes from comment #7) > (In reply to Zeeshan Ali (Khattak) from comment #6) > > (In reply to vladimir benes from comment #5) > > > (In reply to Zeeshan Ali (Khattak) from comment #4) > > > > * Can you manually reproduce it? > > > > > > > yes > > > > > > > * Please try to reproduce on baremetal. > > > > > > > will do > > > > > > > * What exactly happens? You create a live box and it fails to start? > > > Box is shutted down so it disappears. Can this be a liveCD issue? I've tried > > > a bunch of them (version 5/6 of tiny core Linux) > > > > From all the input so far, I'd say it's like a qemu crash. Might be related > > to nested kvm so that's why I asked if you can produce on baremetal. > > Actually I cannot reproduce on bare. I've hit several more crashes but not > this one. So nested KVM, very likely. Ok, good we are getting somewhere. If you can ensure that Qemu actually crashes, could you please file a bug against qemu for this? > not sure if related to fedora so filed there: > https://bugzilla.redhat.com/show_bug.cgi?id=1247941 > https://bugzilla.redhat.com/show_bug.cgi?id=1247945 Would be very helpful if you could try reproducing it against git master.
it's still here it's nested KVM and I see an error in journal, so it looks like libvirt/qemu issue Nov 25 19:28:55 localhost.localdomain libvirtd[18841]: internal error: End of file from monitor
(In reply to vladimir benes from comment #9) > it's still here > > it's nested KVM and I see an error in journal, so it looks like libvirt/qemu > issue > Nov 25 19:28:55 localhost.localdomain libvirtd[18841]: internal error: End > of file from monitor Is it possible that Qemu is crashing?
-- GitLab Migration Automatic Message -- This bug has been migrated to GNOME's GitLab instance and has been closed from further activity. You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.gnome.org/GNOME/gnome-boxes/issues/65.