On Thu, Sep 26, 2013 at 11:53:30AM +0200, Cédric Bosdonnat wrote:
The bug is that after shutting down an lxc full-distro container,
start
always fails. It turns out that shutdown doesn't clean the machine on
the machined side. A workaround is to machinectl terminate the
container. A proper fix would be that machinectl tracks the state of the
init process and cleans the machine if that one disappears.
This fix is the reasonable fix we can have on the libvirt side: really
try to start the container manually if starting it with systemd failed.
It currently never calls the virCgroupNewMachineManual function.
---
src/util/vircgroup.c | 3 ---
1 file changed, 3 deletions(-)
diff --git a/src/util/vircgroup.c b/src/util/vircgroup.c
index e99caf5..466c170 100644
--- a/src/util/vircgroup.c
+++ b/src/util/vircgroup.c
@@ -1656,9 +1656,6 @@ virCgroupNewMachine(const char *name,
group)) == 0)
return 0;
- if (rv == -1)
- return -1;
-
return virCgroupNewMachineManual(name,
drivername,
pidleader,
NACK, ignoring failure to create the machine via systemd is not
acceptable. Either systemd must be fixed, or we should delete the
bits systemd is leaving behind ourselves, or both.
Daniel
--
|:
http://berrange.com -o-
http://www.flickr.com/photos/dberrange/ :|
|:
http://libvirt.org -o-
http://virt-manager.org :|
|:
http://autobuild.org -o-
http://search.cpan.org/~danberr/ :|
|:
http://entangle-photo.org -o-
http://live.gnome.org/gtk-vnc :|