On a Jenkins machine I would like to create a docker container with a specified name only if it does not already exist (in a shell script). I thought I might run the command to create the container regardless and ignore the failure if there was one, but this causes my jenkins job to fail.
Hence, I would like to know how I can check if a docker container exists or not using bash.
You can check for non-existence of a running container by grepping for a <name> and fire it up later on like this:
[ ! "$(docker ps -a | grep <name>)" ] && docker run -d --name <name> <image>
Better:
Make use of https://docs.docker.com/engine/reference/commandline/ps/ and check if an exited container blocks, so you can remove it first prior to run the container:
if [ ! "$(docker ps -a -q -f name=<name>)" ]; then
if [ "$(docker ps -aq -f status=exited -f name=<name>)" ]; then
# cleanup
docker rm <name>
fi
# run your container
docker run -d --name <name> my-docker-image
fi
I suppose
docker container inspect <container-name> || docker run...
since docker container inspect call will set $? to 1 if container does not exist (cannot inspect) but to 0 if it does exist (this respects stopped containers). So the run command will just be called in case container does not exist as expected.
You can use filter and format options for docker ps command to avoid piping with unix utilities like grep, awk etc.
name='nginx'
[[ $(docker ps --filter "name=^/$name$" --format '{{.Names}}') == $name ]] ||
docker run -d --name mynginx <nginx-image>
Just prefix the name with ^/ and suffix with $. It seems that it is a regular expression:
CONTAINER_NAME='mycontainername'
CID=$(docker ps -q -f status=running -f name=^/${CONTAINER_NAME}$)
if [ ! "${CID}" ]; then
echo "Container doesn't exist"
fi
unset CID
Even shorter with docker top:
docker top <name> || docker run --name <name> <image>
docker top returns non-zero when there are no containers matching the name running, else it returns the pid, user, running time and command.
Robust grep ^$ without undocumented behavior
This is the most precise and flexible approach I could find:
container_name=mycont
if sudo docker ps -a --format '{{.Names}}' | grep -Eq "^${container_name}\$"; then
echo exists
else
echo 'does not exist'
fi
Rationale:
docker container inspect and that -f name takes regular expressions of some kind.Python
A Python version for convenience since I ended up using it in a project:
containers = subprocess.check_output([
'sudo',
'docker',
'ps',
'-a',
'--format', '{{.Names}}',
]).decode()
if container_name in containers.split():
# Exists.
This is my solution to check for a running container, stop it, then remove it.
CNAME=$CONTAINER_NAME-$CODE_VERSION
if [ "$(docker ps -qa -f name=$CNAME)" ]; then
echo ":: Found container - $CNAME"
if [ "$(docker ps -q -f name=$CNAME)" ]; then
echo ":: Stopping running container - $CNAME"
docker stop $CNAME;
fi
echo ":: Removing stopped container - $CNAME"
docker rm $CNAME;
fi
I've had to search this too many times because even the 100+ answer above doesn't actually work. I think the reason is a misunderstanding on docker ps. docker ps lists RUNNING containers. docker ps -q does the same but the output is striped to include only the container_id. docker ps -a lists ALL containers (running or not). docker ps -qa then is a simple list of all containers while docker ps -q is a simple list of running containers. docker ps -q -f name=ContainerName is then a simple list of running containers with the name ContainerName. docker ps -qa -f would include exited containers as well so the logic must be to check -a (there, running or not), then without -a to see if it's not only there, but running (and needs to be stopped first).
I use following code to determine if docker container exists:
CONTAINER_NAME='my_docker_container'
# Checking if docker container with $CONTAINER_NAME name exists.
COUNT=$(docker ps -a | grep "$CONTAINER_NAME" | wc -l)
if (($COUNT > 0)); then
echo 'container exists'
fi
-a if you want to list all states (stopped, running etc.)
-f status=running (no need if you want to list only running containers)
$ docker ps -f name=containerName | grep containerName
If there is 1 or more lines status code of command will be 0. If there is no container, grep will be ended with 1 status code. You can handle it by:
Windows
$ echo %errorlevel%
Linux
$ echo $?
Ref: https://www.cyberciti.biz/faq/linux-bash-exit-status-set-exit-statusin-bash/
if [[ $(sudo docker inspect --format . <container-name>) == "." ]]; then
docker run <container-name>;
fi
Explanation:
There is a similar response already. The difference here is the --format . option (you can also use -f .). This removes all the details from the inspect command. Docker uses the go template format, which in this case means that it will copy to the output anything it does not recognize.
So -f itIsThere will return itIsThere if a container with that namex exits. If it doesn't, docker will return an error code and message (Error: No such object: <container-name>).
I found this one in Jenkins logs.
Here an example of a docker container existence check by exact name:
(docker ps -a --format {{.Names}} | grep container_name -w) && echo Exists || echo "Doesn't exist"
to skip Exited container:-
CONTAINER_NAME=xyz
if [ ! "$(docker ps -a | grep ${CONTAINER_NAME} | grep -v Exited)" ]; then echo "${CONTAINER_NAME} is not running"; else echo "${CONTAINER_NAME} is running"; fi
Piggy-backing off of @Fernando César.
This checks for the container id from the given container name and suppresses error output ( 2> /dev/null ).
CONTAINER_NAME="awesome-container"
CONTAINER_ID=$(docker inspect --format="{{.Id}}" ${CONTAINER_NAME} 2> /dev/null)
if [[ "${CONTAINER_ID}" ]]; then
# container found.
else
# container not found.
fi
In bash script I check if container exists by name like this :
CONTAINER_NAME="my-container-name"
if ! docker container ls -a | grep -Fq "$CONTAINER_NAME" 1>/dev/null; then
echo "could not found container $CONTAINER_NAME..."
fi
