On your host (which is a raring qemu VM, bridged to hardware nic, on a precise host) if I do
while [ 1 ]; do
mv -f outout outout2
lxc-execute -n r1 -l info -o outout -- ps
done
after some time, lxc-execute hangs. But the debug output file shows both the good and bad runs made it to at least the very last debug statement.
On my raring laptop I cannot reproduce that. That suggests this is a race triggered when the kernel is slower to do <something> with the veth nics on cleanup.
On your host (which is a raring qemu VM, bridged to hardware nic, on a precise host) if I do
while [ 1 ]; do
mv -f outout outout2
lxc-execute -n r1 -l info -o outout -- ps
done
after some time, lxc-execute hangs. But the debug output file shows both the good and bad runs made it to at least the very last debug statement.
On my raring laptop I cannot reproduce that. That suggests this is a race triggered when the kernel is slower to do <something> with the veth nics on cleanup.
Still looking...