Hi Sean, On Wed, May 20, 2015 at 03:14:43PM -0400, Sean McNamara wrote: > I am using an Ubuntu 14.04 host (upstart init) running linux-generic amd64 > kernel, with lxd-spawned unprivileged lxc container running 15.04 (systemd > init). Latest 15.04 image with no custom software inside the container, and > nothing installed but the barebones. > > I applied a profile specifying a custom bridge device and static IP; this > config works fine for multiple 14.04 containers. Each of these 14.04 > containers can be successfully started and stopped at will. > > But when I create a 15.04 amd64 guest, it will start successfully and is > usable, but it won't *stop*. Even trying to kill the container processes > from the host doesn't seem to get rid of it. The container just keeps > running unless I reboot the system. `lxc stop name` just hangs forever. > There are only systemd processes and /bin/bash running in the container. > > Should the lxc command have a '-f' flag to send SIGKILL to all the child > processes of a container in case one of them is hung?
You can do lxc stop $container --force, which does effectively this. > There is nothing useful in the lxc or lxd logs. All it says is that the > container changed to RUNNING state. Nothing about it changing to any > stop/stopping state. > > P.S. - I looked in dmesg, and none of the "stuck" processes have generated > any kernel oops/warn/etc. I've seen this in some cases too, but not been able to reproduce it. I thing stgraber looked at it a while ago, cc-ing him here to see what he knows. Tycho > Thanks, > > Sean > _______________________________________________ > lxc-users mailing list > [email protected] > http://lists.linuxcontainers.org/listinfo/lxc-users _______________________________________________ lxc-users mailing list [email protected] http://lists.linuxcontainers.org/listinfo/lxc-users
