Moby: Clarification on how to stop containers with --restart=always

[hamiltont]

What's the recommended way to terminate (permanently) containers that were started with --restart=always? If I kill them and then restart the daemon, they come back. If I kill them, remove that container and it's image, then they stay dead. Is this intended behavior? Is there an easier way to say "kill and don't restart"?

[cpuguy83]

I'm not sure if it's intended behavior to restart a stopped container on daemon restart... but for sure docker rm <container> would be all that is needed, no need to remove the image.

[88plug]

docker stop $(docker ps -a -q) &
docker update --restart=no $(docker ps -a -q) &
systemctl restart docker

[zrml]

The command above does not work for me with 17.06 CE

$ docker stop $(docker ps -a -q) &
[1] 10347
$ docker update --restart=no $(docker ps -a -q) &
[2] 10349
$ sudo systemctl restart docker95ca5f77f47f
e4729fa5a4eb
4145ccf79320
95ca5f77f47f
e4729fa5a4eb
4145ccf79320

[1]- Done docker stop $(docker ps -a -q)
[2]+ Done docker update --restart=no $(docker ps -a -q)

$ docker ps
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
4f5b57caaff3 quay.io/prometheus/prometheus:latest "/bin/prometheus -..." 11 seconds ago Up 5 seconds 9090/tcp monitoring_prometheus.1.92mo1fvdkzk19vszcqrlmlm5x

Any other suggestions please?
Thanks

[cpuguy83]

@zrml If you call docker stop on a container, the container will not be restarted unless you restart the daemon (and it has --restart=always)

[zrml]

@cpuguy83 sorry I'm late on this (too much traveling).
The container has --restart=always.
I am able to stop it, but shortly afterwards the daemon restarts it again.
The daemon does not have a --restart=always policy in /lib/systemd/system/docker.service.

Where is that flag for that container maintained? Can I get to it and reset it to --restart=no anywhere?

Thanks

[thaJeztah]

@zrml the --restart=always is set on docker run. If the container you're stopping is part of a swarm service though, docker will start a new container to replace it; that's expected

[zrml]

@thaJeztah , you stand correct, it was set at docker run event. This is not part of a swarm. Thanks

[vharsh]

Helpful
https://stackoverflow.com/questions/27283131/how-can-i-stop-and-delete-a-docker-container-launched-with-restart-always-option

[thaJeztah]

You can also update the restart policy of the container to prevent it from being restarted; see the documentation for docker update;

docker update --restart=no my-container

[andrewb-ms]

The restart policy suggestion doesn't work. This seems really broken - there needs to be a "docker kill" command that shuts down the container and nukes it from the restart policy.

[cpuguy83]

@andrewb-ms --restart=unless-stopped

[lucsrods]

None of the solutions here worked for me, but after I verified the services (docker service ls) and removed them (docker service rm <service-id>), I was finally able to stop the containers.

I'm on Docker version 17.12.0-ce

[thaJeztah]

@lucasar that's a different situation (not related to --restart=always), and the intended behaviour; if you start a swarm service, docker swarm reconciles the service's actual state with the desired state. That means that if you told it to start (e.g.) 5 instances of the service, and an instance of that service goes down (whatever the reason), Docker will create a new instance to replace the one that failed.

[lucsrods]

@thaJeztah Oh ok, got it, thanks for the clarification. :smile:

[azizzoaib786]

docker service rm <service-id> works.. and followed by docker stop $(docker ps -a -q) & docker update --restart=no $(docker ps -a -q)

[ohcibi]

I had two containers that were created with docker compose (not sure if docker-compose is related actually) and I was not able to stop the containers, they kept coming up. I tried to update with restart=no, restart=unless-stopped and restart=never (the latter gave an error as never is not supported) but none of that helped. Even removing the containers did not help at all. Running Docker version 18.03.1-ce, build 9ee9f40

[langzixdj]

You can also update the restart policy of the container to prevent it from being restarted; see the documentation for docker update;

docker update --restart=no my-container

thx Sloved

[gonvaled]

Stopping a container with restart=always is inconsistent:

• containers which have managed to start normally will indeed be stopped (until manually restarted, or the daemon is restarted)
• containers which are stuck in a restart loop (because they are failing to start-up) will NOT be stopped

More details here: https://stackoverflow.com/q/53003404/647991

[thaJeztah]

containers which are stuck in a restart loop (because they are failing to start-up) will NOT be stopped

I suspect for those, that the restart process is already triggered (with a delay), so after the delay, a new start will be sent to those containers

[gonvaled]

containers which are stuck in a restart loop (because they are failing to start-up) will NOT be stopped

I suspect for those, that the restart process is already triggered (with a delay), so after the delay, a new start will be sent to those containers

Probably. From a user point of view, it is confusing: you stop a container and sometimes it stops and sometimes it doesn't.

[teevi]

What about docker rm -f container_name ?

[CatBia]

What about docker rm -f container_name ?

No, the container will not be removed and the restar-docker-hell will be start over and over and over.
If you're in Windows, just stop the Docker Swarm docker swarm leave --force, and, after that, stop them all.

[thaJeztah]

No, the container will not be removed and the restar-docker-hell will be start over and over and over.
If you're in Windows, just stop the Docker Swarm docker swarm leave --force, and, after that, stop them all

That looks unrelated to this issue; if a container is part of a swarm service, stopping or removing the container will cause the swarm reconciliation loop to kick in and spin up a new container to replace it. To remove running containers of a swarm service, remove the service itself. (see https://github.com/moby/moby/issues/23704)

[Bill0412]

@zrml the --restart=always is set on docker run. If the container you're stopping is part of a swarm service though, docker will start a new container to replace it; that's expected

Yes, my docker container was inside a stack. So for me to stop my docker, instead of running docker stop <docker_name>, I had to use docker stack rm <stack_name>.

[vege10]

My docker container status shows UP but when I dso to container, the processes are Stopped.
Based on the samson.stdout log, can see that it always "Restarting samson".

[vege10]

Hello from SAMSON after detach: 21757
3-Mar-20 00:28:16.715 Samson will listen on port 40554 for local processes
3-Mar-20 00:28:16.715 Read state as 'running'
3-Mar-20 00:28:16.966 Starting personality.sh [21800] OK
3-Mar-20 00:28:17.004 Starting pfwdog.sh [21805] OK
3-Mar-20 00:28:17.039 Starting splinter [MONDO] OK
3-Mar-20 00:28:17.040 Starting cvp_server [MONDO] OK
3-Mar-20 00:28:17.041 Starting clam_server [MONDO] OK
3-Mar-20 00:28:17.042 Starting dgi_server [MONDO] Tue Mar 3 00:28:21 +08 2020 Samson has stopped: 0
Tue Mar 3 00:28:22 +08 2020 Restarting samson
Hello from SAMSON after detach: 22004
3-Mar-20 00:28:22.280 Samson will listen on port 42151 for local processes
3-Mar-20 00:28:22.280 Read state as 'running'
3-Mar-20 00:28:22.531 Starting personality.sh [22047] OK
3-Mar-20 00:28:22.570 Starting pfwdog.sh [22052] OK
3-Mar-20 00:28:22.606 Starting splinter [MONDO] OK
3-Mar-20 00:28:22.607 Starting cvp_server [MONDO] OK
3-Mar-20 00:28:22.608 Starting clam_server [MONDO] OK
3-Mar-20 00:28:22.609 Starting dgi_server [MONDO] Tue Mar 3 00:28:26 +08 2020 Samson has stopped: 0
Tue Mar 3 00:28:27 +08 2020 Restarting samson

[perfecto25]

the only thing that worked for me to kill these zombie containers was running a one liner,

sets restart=no, kills container, removes image w force

docker ps -a --format="{{.ID}}" | xargs docker update --restart=no | xargs docker stop | xargs docker rmi $(docker images -qa) --force