I have two Docker containers: node-a, node-b. One of them (node-b) should send http request to other (node-a). I'm starting them with Docker Compose. When I'm trying to up them with Compose I face an error:
Get http://node-a:9098: dial tcp 172.18.0.3:9098: getsockopt: connection refused
EXPOSE is declared in Docker file of a-node:
EXPOSE 9098
docker-compose.yml:
version: '3'
services:
node-a:
image: a
ports:
- 9098:9098
volumes:
- ./:/a-src
depends_on:
- redis
node-b:
image: b
volumes:
- ./:/b-src
depends_on:
- node-a
Forwarding is enabled. I believe a server starts because it works well without Docker.
Where I should pay attention? What could cause a problem?
EDIT:
I've tried to add links but it had no effect:
node-b:
image: b
volumes:
- ./:/b-src
links:
- node-a
depends_on:
- node-a
Also links seemed to be deprecated and does the same thing as depends_on in 2+ version of docker-compose.yml:
docker-compose execute V2 files, it will automatically build a network between all of the containers defined in the file, and every container will be immediately able to refer to the others just using the names defined in the docker-compose.yml file.
Link a container to the service using links. (docker-compose documentation on links).
Example:
node-b:
image: b
volumes:
- ./:/b-src
depends_on:
- node-a
links:
- node-a
Related
Firstly, I'd like to say that I'm not a server admin. I'm a web programmer tasked with setting up a development server and I have no idea what I'm doing. I may not be doing things according to best practice or the way you might do them. Unfortunately, with Traefik, there are 3 ways to do everything and so 2/3 of the answers that I've come across aren't compatible with my implementation and I can't figure out how to make them work. Furthermore, this isn't my only (or even primary) job duty.
Here's the setup:
Single-server docker environment on a Linode server with Ubuntu 20.04
I have one stack with Traefik, Traefik Hub, Portainer, and WhoAmI configured and working (mostly) correctly. I don't have the DNS challenge working right with Let's Encrypt, but I don't really care about that at this point. I don't really need a wildcard certificate.
I created a mariadb container. We're mostly a WordPress shop and I'd like to have one container for all the databases we work with rather than configuring a database on an environment-by-environment basis.
I created an external bridge network, named "maverick-net" and all of the stacks are connected to it.
I have a self-hosted GitHub runner listening for changes to the "dev" branch of the project. The runner pulls down the latest repo, writes GitHub secrets to a local .env file, runs composer install and then docker-compose up -d. (That's the reason behind the obscenely-long bind mount paths.)
I'm trying to make the code for these WordPress projects reusable as much as possible, so there's a lot of .env variables in the different files. At some point I'll probably move those over to docker secrets, but at this point it's a development server and not as critical.
My issue is that I haven't been able to get a WordPress site up and running, and I keep hitting a "Bad Gateway" error. When I curl the URL from inside the traefik container, I get... wait for it... "Bad Gateway."
Clearly there's something I'm missing, but I've been slamming my head against a brick wall for weeks trying different approaches to get this running and I need help. There has to be something I'm not getting about docker networks in general because my wp-cli container never has been able to connect to the database, regardless of whether I start it in the same stack or if i try to connect to it on the maverick-net network.
My traefik stack (side note, I'd really like to split these command entries into static and dynamic config files, but that's a task for another day):
version: "3.9"
secrets:
linode_token:
file: "../secrets/linode_token.secret"
services:
traefik:
container_name: traefik
image: "traefik:latest"
command:
- --log.level=DEBUG
- --log.filePath=./traefik.log
- --accessLog=true
- --accessLog.filePath=./access.log
- --accessLog.bufferingSize=100
- --accessLog.filters.statusCodes=400-499
- --api
- --api.dashboard=true
- --api.insecure=false
- --entrypoints.web.address=:80
- --entrypoints.websecure.address=:443
- --providers.docker
- --providers.docker.watch=true
- --providers.docker.exposedbydefault=false
- --certificatesresolvers.leresolver.acme.dnsChallenge=true
- --certificatesresolvers.leresolver.acme.dnsChallenge.provider=linodev4
- --certificatesresolvers.leresolver.acme.httpchallenge=true
- --certificatesresolvers.leresolver.acme.httpchallenge.entrypoint=web
- --certificatesresolvers.leresolver.acme.email=xxxxxxxxxxx#xxxxxxxxx.xxx
- --certificatesresolvers.leresolver.acme.storage=./acme.json
#- --certificatesresolvers.leresolver.acme.caserver=https://acme-staging-v02.api.letsencrypt.org/directory
- --certificatesresolvers.leresolver.acme.caserver=https://acme-v02.api.letsencrypt.org/directory
- --experimental.hub=true
- --hub.tls.insecure=true
- --metrics.prometheus.addrouterslabels=true
ports:
- "80:80"
- "443:443"
volumes:
- /var/run/docker.sock:/var/run/docker.sock:ro
- ~/certs-data/acme.json:/data/letsencrypt/acme.json
- ./static.yml:/static.yml:ro
- ./configs:/configs
- ~/certs-data/:/data/letsencrypt/
secrets:
- "linode_token"
environment:
TZ: America/Chicago
LINODE_TOKEN_FILE: "/run/secrets/linode_token"
labels:
- "traefik.enable=true"
- "traefik.docker.network=maverick-net"
- "traefik.http.routers.http-catchall.rule=hostregexp(`{host:.+}`)"
- "traefik.http.routers.http-catchall.entrypoints=web"
- "traefik.http.routers.http-catchall.middlewares=redirect-to-https"
- "traefik.http.middlewares.redirect-to-https.redirectscheme.scheme=https"
- "traefik.http.routers.traefik.tls.certresolver=leresolver"
- "traefik.http.routers.traefik.rule=Host(`XXXXX.XXXXXXXXXX.XXX`)"
- "traefik.http.routers.traefik.entrypoints=websecure"
- "traefik.http.routers.traefik.service=api#internal"
- "traefik.http.routers.traefik.middlewares=traefik-auth"
- "traefik.http.middlewares.traefik-auth.basicauth.users=XXXX:$$apr1$$XXXXX$$XXXXXXXXXXXXXXX"
- "traefik.http.routers.api.entrypoints=websecure"
networks:
- maverick-net
hub-agent:
image: ghcr.io/traefik/hub-agent-traefik:experimental
pull_policy: always
container_name: hub-agent
restart: on-failure
command:
- run
- --hub.token=XXXXXXX-XXXX-XXXX-XXXX-XXXXXXXXXX
- --auth-server.advertise-url=http://hub-agent
- --traefik.host=traefik
- --traefik.tls.insecure=true
volumes:
- /var/run/docker.sock:/var/run/docker.sock:ro
depends_on:
- traefik
networks:
- maverick-net
portainer:
image: portainer/portainer-ce:latest
command: -H unix:///var/run/docker.sock
container_name: portainer
restart: always
volumes:
- /var/run/docker.sock:/var/run/docker.sock
- portainer_data:/data
labels:
# Frontend
- "traefik.enable=true"
- "traefik.http.routers.frontend.rule=Host(`XXXXX.XXXXXXXXXX.XXX`)"
- "traefik.http.routers.frontend.entrypoints=websecure"
- "traefik.http.services.frontend.loadbalancer.server.port=9000"
- "traefik.http.routers.frontend.service=frontend"
- "traefik.http.routers.frontend.tls.certresolver=leresolver"
networks:
- maverick-net
whoami:
image: "traefik/whoami"
container_name: "whoami"
labels:
- "traefik.enable=true"
- "traefik.http.routers.whoami.rule=Host(`XXXXX.XXXXXXXXXX.XXX`)"
- "traefik.http.routers.whoami.entrypoints=websecure"
- "traefik.http.routers.whoami.tls.certresolver=leresolver"
networks:
- maverick-net
volumes:
portainer_data:
networks:
maverick-net:
external: true
My mariadb stack:
version: "3"
networks:
# enable connection with Traefik
maverick-net:
external: true
services:
mariadb:
container_name: mariadb
image: mariadb:10.7
restart: always
volumes:
- "/home/xxxxxxxxxx/docker/mariadb/data:/var/lib/mysql"
expose:
- "3306"
env_file: .env
environment:
MYSQL_ROOT_PASSWORD: ${MYSQL_ROOT_PWD}
MYSQL_USER: ${ADMIN_DB_USER}
MYSQL_PASSWORD: ${ADMIN_DB_PWD}
networks:
- maverick-net
And finally my WordPress stack:
version: '3.8'
networks:
maverick-net:
external: true
# volumes:
# db_data:
services:
# mariadb:
# container_name: ${WORDPRESS_DB_NAME}-db
# image: mariadb:10.7
# restart: always
# volumes:
# - "db_data:/var/lib/mysql"
# env_file: .env
# environment:
# MYSQL_ROOT_PASSWORD: ${MARIADB_ROOT_PASSWORD}
# MYSQL_USER: ${ADMIN_DB_USER}
# MYSQL_PASSWORD: ${ADMIN_DB_PWD}
wordpress:
container_name: ${WORDPRESS_DB_NAME}-wp
image: wordpress:6.0.2-fpm
volumes:
- type: bind
source: ${PROJECT_ROOT}/${WORDPRESS_DB_NAME}/${PROJECT_NAME}/${PROJECT_NAME}/wp
target: /var/www/html
restart: always
env_file: .env
environment:
WORDPRESS_DB_HOST: mariadb
MARIADB_ROOT_PASSWORD: ${MARIADB_ROOT_PASSWORD}
WORDPRESS_DATABASE_USER: ${WORDPRESS_DB_USER}
WORDPRESS_DATABASE_PASSWORD: ${WORDPRESS_DB_PASSWORD}
WORDPRESS_DATABASE_NAME: ${WORDPRESS_DB_NAME}
labels:
# The labels are useful for Traefik only
- "traefik.enable=true"
- "traefik.docker.network=maverick-net"
# Get the routes from https
- "traefik.http.routers.${WORDPRESS_DB_NAME}.rule=Host(`${DEV_URL}`)"
- "traefik.http.routers.${WORDPRESS_DB_NAME}.entrypoints=websecure"
- "traefik.http.routers.${WORDPRESS_DB_NAME}.tls.certresolver=leresolver"
networks:
- maverick-net
wordpress-cli:
container_name: ${WORDPRESS_DB_NAME}-cli
image: wordpress:cli
volumes:
- type: bind
source: ${PROJECT_ROOT}/${WORDPRESS_DB_NAME}/${PROJECT_NAME}/${PROJECT_NAME}/wp
target: /var/www/html
env_file: .env
environment:
WORDPRESS_DB_HOST: mariadb
MARIADB_ROOT_PASSWORD: ${MARIADB_ROOT_PASSWORD}
WORDPRESS_DATABASE_USER: ${WORDPRESS_DB_USER}
WORDPRESS_DATABASE_PASSWORD: ${WORDPRESS_DB_PASSWORD}
WORDPRESS_DATABASE_NAME: ${WORDPRESS_DB_NAME}
networks:
- maverick-net
depends_on:
- wordpress
As far as I know, you can connect to containers in the same network by using their service name.
So for example you are trying to curl to the Wordpress container from the Traefik Container.
curl 'http://wordpress/'
Should work.
In another project I use an nginx container with php-fpm.
I need to send my curl requests to the nginx container, because the php-fpm container does not handle server requests directly:
// does not work
curl 'http://php-debug/index.html'
// result
curl: (7) Failed to connect to php-debug port 80: Connection refused
// https also does not work
curl 'https://php-fpm/index.html'
// result
curl: (7) Failed to connect to php-fpm port 443: Connection refused
// This does work
curl 'http://nginx/index.html'
// result
<HTML...
For some reason https: curl requests fail, but on http I get the correct result so for local development I think it's ok.
If you are interested in a more managed solution, you can check out warden.dev. It contains a template for Wordpress too (which I use succesfully for local development). I have been using this exclusively. If you have questions how to setup WP CLI on this solution, feel free to contact me.
It comes with portainer, traefik, ssl and dns and mailhog.
Configuration is pretty straight forward, I can set up a new project within an hour and connect to the database and containers in my IDE.
https://docs.warden.dev/environments/types.html#wordpress
I am trying to connect to a database that has an IP of x.x.x.x from my Docker container
Getting this error
java.net.NoRouteToHostException: No route to host (Host unreachable)
Tried running container using --network=host which has a similar approach to the above attempt
As I mentioned in the comments, here is the sample docker-compose file.
version: '3.7'
services:
entitygraph:
image: entitygraph-by-jar:latest
container_name: entitygraph
restart: always
networks:
- eg-net
ports:
- 9999:8080
environment:
SPRING_DATASOURCE_URL: jdbc:mysql://eg-mysql/customers?useSSL=false
SPRING_PROFILES_ACTIVE: mysql
eg-mysql:
image: mysql:5.7
restart: always
networks:
- eg-net
container_name: eg-mysql
environment:
MYSQL_DATABASE: customers
MYSQL_ALLOW_EMPTY_PASSWORD: "yes"
MYSQL_ROOT_PASSWORD:
networks:
eg-net:
name: eg-net
In this file, the application entitygraph is trying to talk to mysql. In my application, the connection string to mysql is as below,
spring.datasource.url=jdbc:mysql://localhost:3306/customers?useSSL=false
So, docker will replace the spring.datasource.url property with the one I specified on my docker-compose file. Note that host:port is eg-mysql, which docker resolves to it's internal IP and will use it to communicate.
I don't know about your application architecture. If I know, I could give you more specific answer to your problem.
I've a question how exactly docker-compose handles environment variables.
services:
wp:
image: wordpress:latest
container_name: "wp"
restart: unless-stopped
links:
- wpdb
environment:
- TZ=Europe/Berlin
- WORDPRESS_DB_HOST=wpdb:3306
- WORDPRESS_DB_USER=wordpress
- WORDPRESS_DB_PASSWORD=password
- WORDPRESS_DB_NAME=wp
volumes:
- ./data:/var/www/html
labels:
- "traefik.enable=true"
- "traefik.backend=wp"
- "traefik.frontend.rule=Host:MASKED"
- "traefik.port=80"
- "traefik.docker.network=web"
networks:
- internal
- web
wpdb:
image: mariadb:latest
restart: unless-stopped
container_name: "wpdb"
environment:
- MYSQL_ROOT_PASSWORD=1234
- MYSQL_USER=wordpress
- MYSQL_PASSWORD=password
- MYSQL_DATABASE=wp
networks:
- internal
labels:
- "traefik.enable=false"
volumes:
- ./sql:/var/lib/mysql
volumes:
data:
sql:
networks:
web:
external: true
internal:
The compose file works great. The containers will be created and work perfectly.
But when I change the defaults at: WORDPRESS_DB_PASSWORD=password and MYSQL_PASSWORD=password.
The Wordpress container throws access denied for user. I also tried to kill the container and volumes.
Hopefully someone has a hint for me.
You should be doing a docker-compose down -v which would delete the named volumes declared in the volumes section. The only downside is that you would be losing all the data created by the service for the first time.
Here is how I could reproduce it -
Used your compose file as reference and on first time used the default password mentioned by you. The services come up fine, I install it and do a Ctrl+C to bring down the service. So all the MYSQL data is written into sql named volume.
When you do a Ctrl+C OR docker-compose down it only removes the containers and networks defined in the service. Not the volumes. Read more about it here
Now when you change password and bring the service back up it still uses the old volumes which has your old password.
So use a docker-compose down -v to remove the volumes too and give it a try.
Here are the steps how I reproduced it
Ctrl+C to stop all the services and then update the docker-compose.yml to update the password and do a docker-compose up again to get access denied error.
Do a docker-compose down -v to clean all the volume too and then do a docker-compose up
On doing a docker-compose down -v you will be losing all the data created by the prior service. Use it cautiously.
I'm new to docker and I'm facing a problem when trying to connect to a container from within another. The wierd stuff is that some containers can indeed be contacted, and others, configured in the same way, not. I tried a zillion fixes found while crawling google, nothing works. I guess this is a simple noob mistake though.
Here's my docker-compose file:
version: '3.4'
services:
mssql.db.posm:
image: "microsoft/mssql-server-linux"
environment:
SA_PASSWORD: "mypassword"
ACCEPT_EULA: "Y"
MSSQL_PID: "Express"
volumes:
- mssql-data:/var/opt/mssql
ports:
- "1433:1433"
networks:
posm:
api.posm:
image: ${DOCKER_REGISTRY}posm.api
build:
context: .
dockerfile: Posm.Api/Dockerfile
expose:
- "6869"
ports:
- "6869:80"
networks:
posm:
cloud.subscription:
image: ${DOCKER_REGISTRY}cloud.subscription
build:
context: ./Services
dockerfile: Cloud.Subscription/Dockerfile
ports:
- "80"
networks:
posm:
catalogmanager.services.posm:
image: ${DOCKER_REGISTRY}posm.services.catalogmanager
build:
context: ./Services
dockerfile: CatalogManager/Dockerfile
ports:
- "80"
networks:
posm:
aliases:
- services.posm
- catalogmanager
productmanager.services.posm:
image: ${DOCKER_REGISTRY}posm.services.productmanager
build:
context: ./Services
dockerfile: ProductManager/Dockerfile
ports:
- "80"
networks:
posm:
aliases:
- services.posm
- productmanager
localizer.services.posm:
image: ${DOCKER_REGISTRY}posm.services.localizer
build:
context: ./Services
dockerfile: Localizer/Dockerfile
ports:
- "80"
networks:
posm:
aliases:
- services.posm
- localizer
networks:
posm:
volumes:
mssql-data:
So, when I connect to the api.posm, the latter is able to successfully connect to mssql.db.posm and to cloud.subscription. It however fails, with a connection refused error, to connect to 'catalogmanager' and 'productmanager', even though I can resolve their hostname from within api.posm... What is happening?
You are giving the alias services.posm both to 'catalogmanager' and 'productmanager':
posm:
aliases:
- services.posm
- catalogmanager
As you now per the documentation (https://docs.docker.com/compose/compose-file/#aliases), this will give the hostname "services.posm" to the containers in the posm network. This means that you are having 2 times the "services.posm" host in the network, and both exposing the port 80. This will not work. Remove the "services.posm" line on both containers aliases, restart and you will be able of accessing them using the catalogmanager and productmanager hosts.
So, the problem was a stack overflow exception in one of my images that was unhandled by my debugger, and that I therefore did not see.
Concerning your answer, Moreno...
You are giving the alias services.posm both to 'catalogmanager' and
'productmanager'
I had duplicated this post on Docker forums for increasing my chances to get a relevant answer, and someone suggested the same. The reason I was using the same aliases for three different services is that I read somewhere (cannot find it now, of course) that you could tag multiple services with the same alias for discovery purposes. I probably got that wrong but Docker doesn't seem to mind at all anyways.
So, thanks for your time nonetheless, sorry that I bothered for such stupid mishandling of my code.
Bottom line is: DOCKER ROCKS
I'm using Docker Compose to create two containers. One runs an Nginx web server which serves the mydomain.com website, and the second needs to send HTTP requests to the first one (using the mydomain.com domain name).
I don't want to have to check the Nginx container's ip each time I run it and then use docker run --add-host on the second container. My goal is to run docker-compose up and that everything be ready.
I know it's not possible, but what I'm looking for is something in the line of:
# docker-compose.yml
nginx_container:
...
second_container:
extra_hosts:
# This is invalid. extra_hosts only accepts ips.
- "mydomain.com:nginx_container"
You can get a similar result using a configuration like this:
version: "3"
services:
api:
image: node:8.9.3
container_name: foo_api
domainname: api.foo.test
command: 'npm run dev'
links:
- "mongo:mongo.foo.test"
- "redis:redis.foo.test"
volumes:
- .:/app
- /app/node_modules
ports:
- "${PORT}:3000"
- "9229:9229"
depends_on:
- redis
- mongo
networks:
- backend
redis:
image: redis:3
container_name: foo_redis
domainname: redis.foo.test
ports:
- "6379:6379"
networks:
- backend
mongo:
image: mongo:3.6.2
container_name: foo_mongo
domainname: mongo.foo.test
ports:
- "${MONGO_PORT}:27017"
environment:
- MONGO_PORT=${MONGO_PORT}
networks:
- backend
networks:
backend:
driver: bridge