CLI must check if all containers started properly

[riuvshin]

today CLI checks only is api server is available, but sometimes some containers may not start this usually happen due to docker bugs like after docker engine restart some container unable to start and the only way to make it work remove stopped container and restart. There is a lot of other possible scenarios where some container may not start so we must check every single container is it at least running container.

here is command how to do that check with inspect this command return true or false
docker inspect -f {{.State.Running}} $CONTAINER_ID

so all we neeed to do is to iterate across all codenvy containers and check if all is in running state.

here is a command how to get all existed (even stopped) codenvy containers:
docker ps -qa --filter="name=codenvy_*"

but that approach could be better if we will find a way how to do that check by container name because I think we could have situation where container was not created so we will not have container id to check. (Im not sure if that situation is possible)

[riuvshin]

So I think algorithm should be like that:

for each image from images file do:

1. get container_id=$(docker ps -aq --filter=ancestor=$IMAGE_NAME)
2. do check is $container_id is not empty, if it is empty we fail which means container didn't started due to some reason
3. do check is container running docker inspect -f {{.State.Running}} $container_id we ok if this return true, fail if false.

@benoitf wdyt?

[benoitf]

yeah I've understood the goal

[benoitf]

@riuvshin
so after analyzing the issue I think your issue is only when docker compose is not able to be started the first time.
If docker compose successfully starts, then it is automatically restarting the containers
and if che doesn't boot in the given amount of time, then we exit with an error

so I think we only need to check docker compose boot sequence. For now if there is a failure it silently exits

[benoitf]

#3731