feature request: high availability mode

[pmlopes]

Sometimes I would like to have a simple HA container, that in case it terminates with a status code other than 0 would relaunch with the same args it used before.

Imagine that you have a python/nodejs app and there was an exception, the node/python process dies and if the flag ha is true then fig would issue again a up command.

This would be a very simple HA feature but good enough for most of deployment scenarios...

[bfirsh]

Related: moby/moby#26

[d11wtq]

Personally I think this is not the responsibility of fig. Fig is an orchestration tool, not a monitoring tool. Just my $0.02. In particular, fig doesn't keep running. It just hits endpoints in the docker API and then returns. I like that. I'd prefer to see this sort of thing added to docker itself, although running any one of the multitude of available process monitors out there (God, Monit, Supervisor, Bluepill) inside the container is quite feasible.

[pmlopes]

sorry for the late reply, if i get it right docker should only containerize a process starting multiple processes is possible that that is the the main idea. What i meant with HA was not really monitoring in the sense of hows the process doing but just simple start and stop events which could trigger running some scripts...

I think this would be very useful for most deployed applications...

[dnephin]

restart is now an option http://docs.docker.com/compose/yml/#working-dir-entrypoint-user-hostname-domainname-mac-address-mem-limit-memswap-limit-privileged-restart-stdin-open-tty-cpu-shares-cpuset-read-only-volume-driver