Age | Commit message (Collapse) | Author |
|
|
|
In starting containers in lxc, I was seeing errors like:
/proc/self/fd/9: 24: kill: No such process
Which indicated the sleep pid had already died.
I'm not sure how or why it was dead, but this just is less annoying in that
case.
|
|
|
|
LP: #1015223
|
|
This changes the way that we avoid cloud-init-nonet hanging in a container.
Previously, under LP: #800824 we tried 'start networking', but that caused
issues described in LP: #1031065.
Here, we emit the net-device-added for any devices that have not yet been
seen.
LP: #1031065
|
|
LP: #861866
|
|
|
|
This should not happen any time in the near future, but /var/run
is actually legacy, so accept that it might not be there.
|
|
|
|
This continues the change in this file that intended to wait
for all networking to be up. The logic that was there would
cause it to start cloud-init immediately if a single non-lo
interface was up.
This will basically just check if 'static-network-up' has occurred
during this boot.
There could be an issue if /var/run was populated from a previous
boot, but since its really expected to be a tmpfs, can't have anything
in it.
|
|
LP: #810044
|
|
LP: #800824
|
|
|
|
to console on missing file
|
|
LP: #714807
|