How do I define the name of image built with docker-compose
Solution 1:
For docker-compose version 2 file format, you can build and tag an image for one service and then use that same built image for another service.
For my case, I want to set up an elasticsearch cluster with 2 nodes, they both need to use the same image, but configured to run differently. I also want to build my own custom elasticsearch image from my own Dockerfile. So this is what I did (docker-compose.yml
):
version: '2'
services:
es-master:
build: ./elasticsearch
image: porter/elasticsearch
ports:
- "9200:9200"
container_name: es_master
es-node:
image: porter/elasticsearch
depends_on:
- es-master
ports:
- "9200"
command: elasticsearch --discovery.zen.ping.unicast.hosts=es_master
You can see that in the first service definition es-master
, I use the build
option to build an image from the Dockerfile in ./elasticsearch
. I tag the image with the name porter/elasticsearch
with the image
option.
Then, I reference this built image in the es-node
service definition with the image
option, and also use a depends_on
to make sure the other container es-master
is built and run first.
Solution 2:
As per docker-compose 1.6.0:
You can now specify both a build and an image key if you're using the new file format.
docker-compose build
will build the image and tag it with the name you've specified, whiledocker-compose pull
will attempt to pull it.
So your docker-compose.yml
would be
version: '2'
services:
wildfly:
build: /path/to/dir/Dockerfile
image: wildfly_server
ports:
- 9990:9990
- 80:8080
To update docker-compose
sudo pip install -U docker-compose==1.6.0
Solution 3:
Option 1: Hinting default image name
The name of the image generated by docker-compose depends on the folder name by default but you can override it by using --project-name
argument:
$ docker-compose --project-name foo build bar
$ docker images foo_bar
Option 2: Specifying image name
Once docker-compose 1.6.0 is out, you may specify build:
and image:
to have an explicit image name (see arulraj.net's answer).
Option 3: Create image from container
A third is to create an image from the container:
$ docker-compose up -d bar
$ docker commit $(docker-compose ps -q bar) foo_bar
$ docker-compose rm -f bar
Solution 4:
Depending on your use case, you can use an image which has already been created and specify it's name in docker-compose
.
We have a production use case where our CI server builds a named Docker image. (docker build -t <specific_image_name> .
). Once the named image is specified, our docker-compose
always builds off of the specific image. This allows a couple of different possibilities:
1- You can ensure that where ever you run your docker-compose
from, you will always be using the latest version of that specific image.
2- You can specify multiple named images in your docker-compose
file and let them be auto-wired through the previous build step.
So, if your image is already built, you can name the image with docker-compose
. Remove build
and specify image:
wildfly:
image: my_custom_wildfly_image
container_name: wildfly_server
ports:
- 9990:9990
- 80:8080
environment:
- MYSQL_HOST=mysql_server
- MONGO_HOST=mongo_server
- ELASTIC_HOST=elasticsearch_server
volumes:
- /Volumes/CaseSensitive/development/wildfly/deployments/:/opt/jboss/wildfly/standalone/deployments/
links:
- mysql:mysql_server
- mongo:mongo_server
- elasticsearch:elasticsearch_server
Solution 5:
According to 3.9 version of Docker compose, you can use image: myapp:tag
to specify name and tag.
version: "3.9"
services:
webapp:
build:
context: .
dockerfile: Dockerfile
image: webapp:tag
Reference: https://docs.docker.com/compose/compose-file/compose-file-v3/