Can’t keep in background on docker
QFJABC opened this issue · 18 comments
Hello,I recently found something strange. The docker container on my Debian server can't keep in the background.I must connect to my server with Putty and it will run. If I exit the terminal,it will stop working.
When I logon the server,and run "docker ps",as you can see,it returns me "up less than a second",which means it just working when I logon!
What command are you using to start the container?
What command are you using to start the container?
docker run -d --name py-kms --restart always -p 1688:1688 ghcr.io/py-kms-organization/py-kms
By the way fix up the picture
Please make sure to pull the latest image once more - and send the output from docker image ls. Because I've started the container with the same command you used (without --name as I've run it before):
$ docker ps
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
502d8f849bdc ghcr.io/py-kms-organization/py-kms "/usr/bin/python3 -u…" 5 hours ago Up 5 hours (healthy) 0.0.0.0:1688->1688/tcp, :::1688->1688/tcp hopeful_cohen
...and it is still running...
Ok,I will try that.
It doesn't work....
I'm sorry, but you failed to pull the image correctly. This is my output:
$ docker pull ghcr.io/py-kms-organization/py-kms
Using default tag: latest
latest: Pulling from py-kms-organization/py-kms
9621f1afde84: Pull complete
6831eb133204: Pull complete
7dc84511a6a9: Pull complete
c212f116beae: Pull complete
6a2cecfb044b: Pull complete
7f1cf1037d65: Pull complete
4f4fb700ef54: Pull complete
Digest: sha256:830c65dac504b2ebdff18e8126f3c66db06860f306a9703e15b598642b8d9de9
Status: Downloaded newer image for ghcr.io/py-kms-organization/py-kms:latest
ghcr.io/py-kms-organization/py-kms:latest
$ docker image ls
REPOSITORY TAG IMAGE ID CREATED SIZE
ghcr.io/py-kms-organization/py-kms latest 3ee7f5ad967e 13 days ago 103MB
As you can see, the latest image hash is 3ee7f5ad967e, while you are using 6b45015e0975. Try to use docker pull again.
Now I pull the image correctly, but the problem still have.
Okay. Then I should be able to reproduce this. Could you please destroy your existing container, recreate it using the run instruction and share the output of docker ps and docker inspect [CONTAINER_NAME]?
Hello, and here are the images.
@simonmicro how can a container get stuck in a startup loop when it's health-check is apparently passing?
In the docker ps output it looks as though the container is successfully daemonised but without a passing health check.
I see in the inspect it's not got a failure streak and also that it's returning zero so what is making docker think it's not started?
Edit:
Looking at the code, I can see we Netcat the KMS port in the container to check it's health. In the inspect output I can see it's returning zero but docker seems to think it's "starting".
@Matthew-Beckett My Windows won't active unless I start Putty for a while.....I don't think it started.
@Matthew-Beckett My Windows won't active unless I start Putty for a while.....I don't think it started.
Is your system dual-stack enabled? I don't expect this to cause an issue but I can see that Docker is binding py-kms to both the IPv4 and IPv6 stack.
Can you try disabling IPv6?
@Matthew-Beckett My Windows won't active unless I start Putty for a while.....I don't think it started.
Can you expand on "unless I start Putty for a while"
What does this mean?
@Matthew-Beckett It means that I must login the system.If I logout the system,it will auto stop.
@simonmicro is this being caused by dual stack being disabled by default in current builds?
@simonmicro is this being caused by dual stack being disabled by default in current builds?
Nope - it should still be active...
According to https://docs.docker.com/engine/reference/builder/#healthcheck the return value of 0 is a success?! I have no idea, why Docker considers that container as failed (health-starting indicates, that maybe the startup-time has not passed -> even upon successful checks the container is still "starting"?)... But IIRC Docker should not act upon that - you would need to use something like https://hub.docker.com/r/willfarrell/autoheal/ for that...
Maybe... I've heard that users could use systemd's user-space services... This means that a user could create a Docker service, which only runs while he is logged into the system and then stopped upon logout... Maybe that's the case here?
@simonmicro Yes, I remember that after docker installed,it asks me to create a Docker service which says "Don't need root to start docker"