How to show the run command of a docker container

Question

I use a third party GUI (Synology Docker package) to setup a docker container. However, its limitation makes me need to run the container from the command line. (I want to map another host ip to bind the port)

Now, since there are lots of settings that already done, I would like to retrieve the original run command that start this container, then I can change the port mapping port to new one. eg. "docker run -p 80:8080 gitlab"

I can't find the way to do so, event use "docker inspect", no such information provided.

Please provide some advice to solve this problem.

Answer 1

At least in version 4.30.0 (Windows) you can use the Copy Docker Run option from the Action menu:

Answer 2

A simpler (?) alternative is to run this docker inspect template, which uses the builtin Go templating capabilities to output a docker run compatible command. The template only covers the most usual command-line options, but it can easily be extended.

This solution has no dependencies on other tools, except docker itself.

docker inspect \
  --format "$(curl -s https://gist.githubusercontent.com/efrecon/8ce9c75d518b6eb863f667442d7bc679/raw/run.tpl)" \
  name_or_id_of_your_running_container

Answer 3

Curious how most answers rely on third parties for that while everything is in the docker inspect JSON. There’s even no need for jq since docker has a JSON formatting option.

Use

docker inspect  -f 'docker run --name {{.Name}} -v {{join .HostConfig.Binds " "}} -p {{.HostConfig.PortBindings}} {{.Config.Image}} {{join .Config.Cmd " "}}' my-osrm-image

E.g. for the actual bash command

docker run -t -i --name my-osrm-image -v "/custom/path/to/data-osrm:/data" -p 80:5000 osrm/osrm-backend osrm-routed --algorithm mld /data/europe-latest.osrm

the above docker inspect command will output

 docker run --name /my-osrm-image -v /custom/path/to/data-osrm:/data -p map[5000/tcp:[{ 80}]] osrm/osrm-backend osrm-routed --algorithm mld /data/europe-latest.osrm

It requires a bit of cleaning but everything is there, and it can be extended to include further options accordingly with the docker inspect JSON output. See also this documentation for the output formatting.

Answer 4

So how to reverse engineering docker run command?

There is a github repository which try to reverse engineering docker run command, but it is not perfect currently, version is 0.1.2. You should follow it for updating. Maybe one day you can use it to get correct run command with it.

$ sudo pip install runlike

# run the ubuntu image
$ docker run -ti ubuntu bash

$ docker ps -a  
# suppose you get the container ID 1dfff2ba0226

# Run runlike to get the docker run command. 
$ runlike 1dfff2ba0226
docker run --name=elated_cray -t ubuntu bash

Github repository: runlike

Updates：

Run without installing (Thanks @tilo)

docker run --rm -v /var/run/docker.sock:/var/run/docker.sock:ro \
    assaflavie/runlike YOUR-CONTAINER

or set alias and put it in your shell's profile

alias runlike="docker run --rm -v /var/run/docker.sock:/var/run/docker.sock:ro assaflavie/runlike"

docker ps

runlike YOUR-CONTAINER

Answer 5

Use following command to get the arguments for all containers:

docker inspect -f "{{.Name}} {{.Path}} {{.Args}}" $(docker ps -a -q)

Answer 6

As noted by @Chris_Lamb, you can use docker inspect to see this and much more. Here is how to load this info to a python dictionary:

import subprocess,json
im_inspect = subprocess.check_output(["docker", "image", "inspect", "datalab-nginx-ui"])
# print(im_inspect.decode("utf-8") ) #if you want to print it
d_inspect = json.loads(im_inspect)
d_inspect
d_inspect[-1]['ContainerConfig']['Cmd']

Example run:

Answer 7

If you do not want to install anything into your current running Docker server setup, you can simply execute (replace $CONTAINER_NAME with the container name you want to have the run arguments):

docker run -it --rm --volume /var/run/docker.sock:/var/run/docker.sock --privileged docker sh -c "apk add --no-cache nodejs nodejs-npm && npm i -g rekcod && rekcod $CONTAINER_NAME"

(for the rekcod method)

or

docker run -it --rm --volume /var/run/docker.sock:/var/run/docker.sock --privileged docker sh -c "apk add --no-cache py-pip && pip install runlike && runlike $CONTAINER_NAME"

(for the runlike method)

Answer 8

All docker files are here you can find cmd and mounts for example ls -la /proc/1 just cat it

cat /proc/1/cmdline
nginx: master process nginx -g daemon off;

Answer 9

I wrote a simple Node-based CLI tool to generate a docker run command from an existing container.

https://www.npmjs.com/package/rekcod

Here's an example:

$ npm i -g rekcod
$ rekcod redis_container

docker run -d --name redis_container --restart always -h a44159e148e1 \
--expose 6379/tcp -e PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin \
-e REDIS_VERSION=3.0.7 -e REDIS_DOWNLOAD_URL=http://download.redis.io/releases/redis-3.0.7.tar.gz \
-e REDIS_DOWNLOAD_SHA1=e56b4b7e033ae8dbf311f9191cf6fdf3ae974d1c \
--entrypoint "/entrypoint.sh" redis "redis-server"

Also handles links and mounted volumes and other stuff.

Not super robust at the moment, but handles more than some of the other things mentioned, and it was more of what I was looking for.

EDIT: In a single command, without installing any software:

docker run --rm -v /var/run/docker.sock:/var/run/docker.sock nexdrew/rekcod redis_container

Answer 10

To reverse the docker run command there is also the following npm package.

https://github.com/nexdrew/rekcod

Answer 11

Use docker inspect:

$ docker inspect foo/bar
[
    {
        # …
        "Config": {
            # …
            "Cmd": [
                "/usr/local/bin/script.sh"
            ],
            # …
        }
    }
]

You can programatically parse this with jq:

$ docker inspect foo/bar | jq -r '.[0]["Config"]["Cmd"][0]'
/usr/local/bin/script.sh

Answer 12

What could be a simpler (robust) option would be to use something like bash-preexec to capture commands that start with "docker run". You could then store these commands somewhere and retrieve them later.

For example, you could add something like this in your bash profile:

[[ -f ~/.bash-preexec.sh ]] && source ~/.bash-preexec.sh
docker_run_history=~/.docker_run_history
docker_clear_history(){
    echo -n > $docker_run_history
}
docker_search_history(){
    search_for="$@"
    [[ -z $search_for ]] && search_for=".*"
    \cat $docker_run_history | grep "$search_for" | tail -1
}
docker_ps_mod(){
    for c in $(docker ps --format "{{.Image}}"); do 
        echo "Container $c was run using:"
        echo -e "\t$(docker_search_history $c)"
    done
}
docker_hook(){
    if [[ $@ =~ ^"docker run".*$ ]]; then
        \echo "$@" >> $docker_run_history 
    fi
}
preexec(){ 
    docker_hook $@
}

Then you could just run your things:

source ~/.bash_profile
docker run -it --rm -v $(pwd)/data:/data -p 8080:80 image
docker run -d daemon
docker_ps_mod

Which outputs:

Container image was run using:
    docker run -it --rm -v $(pwd)/data:/data -p 8080:80 image
Container daemon was run using:
    docker run -d daemon

Answer 13

Currently it seems we have to go 'docker inspect ' and then manually recreate the run command.

I have found someone attempting to write a bash script to do this: https://gist.github.com/miracle2k/c85b7b077fdb8d54bc89

but it is incomplete and depends on jq.

Answer 14

That docker run command isn't specified in the Dockerfile or any other docker-related documents.

Either you find an example in the documentation associated to your container, or you can infer that docker run with (at least for the command and port mapping) a docker ps -a (but that won't give you the possible --volumes-from options)

Check also /usr/syno/etc/packages/Docker-GitLab/config

This differ from the gitlab config itself, which on Synology is available in /usr/syno/etc/packages/Docker/synology_gitlab.config