Docker Cloud is a powerful tool that allows us to automate image builds, provision docker nodes, and create CI/CD pipelines. A month ago Docker introduced swarm mode for the Docker Cloud. And now teams and individuals can create and manage new or register their existing swarms and operate with them with docker id.
Our goal is to create a new swarm and deploy scalable WordPress.
Create a WP swarm
Before we begin we should link our Docker Cloud to Amazon Web Services. After this procedure, we can create swarms with several clicks.
Let's create our WP swarm and choose EC2 instance type and a number of manager and worker nodes below.
P.S. I hope Docker Team will add a possibility to select different instance types for labeled groups of worker nodes
To persist our data we will use EFS and we should remember that only three regions offer efs support in NA.
After clicking the Create button it will take about five minutes to deploy swarm on Amazon. Docker Cloud uses CloudFormation template, the final result looks like this
To get instructions on how to connect to the new swarm just click on it when the status changed to deploy.
I've created swarm with three managers and two worker nodes:
$ docker node ls
ID HOSTNAME STATUS AVAILABILITY MANAGER STATUS
ah7s06zgqh04bq7vjvnlsp115 ip-172-31-36-181.ec2.internal Ready Active Reachable
lq3a5989rsykyofbuxvo9jzal * ip-172-31-2-231.ec2.internal Ready Active Leader
mq31ffl0hj96549rch0b1jj5h ip-172-31-28-121.ec2.internal Ready Active
oe5wa3z87tz3smn8yrezl5y8p ip-172-31-17-95.ec2.internal Ready Active Reachable
zkjm6xgfqubkr91fpe2lrhcfz ip-172-31-0-161.ec2.internal Ready Active
Now let's create temporary wordpress service to test our swarm.
We need secrets for our database:
$ openssl rand -base64 20 | docker secret create root_db_password -
$ openssl rand -base64 20 | docker secret create wp_db_password -
overlay network:
$ docker network create -d overlay wp
and MariaDB service:
$ docker service create \
--name mariadb \
--replicas 1 \
--constraint=node.role==manager \
--network wp \
--secret source=root_db_password,target=root_db_password \
--secret source=wp_db_password,target=wp_db_password \
-e MYSQL_ROOT_PASSWORD_FILE=/run/secrets/root_db_password \
-e MYSQL_PASSWORD_FILE=/run/secrets/wp_db_password \
-e MYSQL_USER=wp \
-e MYSQL_DATABASE=wp \
mariadb:10.1
vx5c0fo9yuucpj50dx9up2xot
$ docker service ps mariadb
ID NAME IMAGE NODE DESIRED STATE CURRENT STATE ERROR PORTS
n8pnarl9x43b mariadb.1 mariadb:10.1 ip-172-31-17-95.ec2.internal Running Running 19 seconds ago
And finally run WordPress:
$ docker service create \
--name wp \
--constraint=node.role==worker \
--replicas 1 \
--network wp \
--publish 80:80 \
--secret source=wp_db_password,target=wp_db_password,mode=0400 \
-e WORDPRESS_DB_USER=wp \
-e WORDPRESS_DB_PASSWORD_FILE=/run/secrets/wp_db_password \
-e WORDPRESS_DB_HOST=mariadb \
-e WORDPRESS_DB_NAME=wp \
wordpress:4.7
n5pruxiwz6ppr5dvqm5a6n33f
Every published port will be automatically added to the ELB entrypoint and we can proceed WP installation using web wizard:
Fix issues
That's simple and great but with this WordPress instance we have several issues:
- No SSL
- No Persistance
- No Scalability (we can scale WordPress service, but wp-content will be different in all containers)
Let's destroy this WP service and try to fix those issues with
$ docker service remove wp mariadb
wp
mariadb
If you deploy swarm to the EFS-compatible region Docker Cloud will install cloudstore:aws storage driver to every node, and this allows you to mount EFS storage as a docker volume. By default this driver uses general purpose EFS. For MariaDB service we will use maxio option. Also we will start traefik service to generate LetsEncrypt SSL and serve our vhost.
Here is our docker-stack.yml:
version: "3.1"services:traefik:
image: traefik:1.2
command: --entryPoints='Name:http Address::80 Redirect.EntryPoint:https' --entryPoints='Name:https Address::443 TLS' --defaultEntryPoints=http,https --acme.entryPoint=https --acme.email=rock@maddevs.io --acme.storage=/etc/traefik/acme/acme.json --acme.domains=maddevs.io --acme.onHostRule=true --docker --docker.swarmmode --docker.domain=maddevs.io --docker.watch
ports:
- 80:80
- 443:443
volumes:
- traefic_vol:/etc/traefik/acme
- /var/run/docker.sock:/var/run/docker.sock
networks:
- traefik
deploy:
placement:
constraints: [node.role == manager]
replicas: 1
update_config:
parallelism: 2
delay: 10s
restart_policy:
condition: on-failurewp:
image: wordpress:latest
volumes:
- wp_content:/var/www/html/wp-content
secrets:
- wp_db_password
environment:
- WORDPRESS_DB_USER=wp
- WORDPRESS_DB_NAME=wp
- WORDPRESS_DB_PASSWORD_FILE=/run/secrets/wp_db_password
- WORDPRESS_DB_HOST=mariadb
networks:
- traefik
- mariadb
deploy:
placement:
constraints: [node.role == worker]
replicas: 1
update_config:
parallelism: 2
delay: 10s
restart_policy:
condition: on-failure
labels:
- "traefik.backend.loadbalancer.swarm=true"
- "traefik.port=80"
- "traefik.frontend.rule=Host:wp.maddevs.io"
- "traefik.docker.network=traefik"mariadb:
image: mariadb
volumes:
- mariadb_vol:/var/lib/mysql
secrets:
- wp_db_password
- root_db_password
environment:
- MYSQL_USER=wp
- MYSQL_DATABASE=wp
- MYSQL_PASSWORD_FILE=/run/secrets/wp_db_password
- MYSQL_ROOT_PASSWORD_FILE=/run/secrets/root_db_password
networks:
- mariadb
deploy:
placement:
constraints: [node.role == manager]
replicas: 1
restart_policy:
condition: on-failuresecrets:
wp_db_password:
external: true
root_db_password:
external: truevolumes:
traefic_vol:
driver: "cloudstor:aws"
wp_content:
driver: "cloudstor:aws"
mariadb_vol:
driver: "cloudstor:aws"
driver_opts:
perfmode: maxionetworks:
traefik:
external: true
mariadb:
external: true
Before start we should create CNAME wp.maddevs.io pointing to our ELB entrypoint. Then create networks and database secrets:
$ openssl rand -base64 20 | docker secret create root_db_password -
$ openssl rand -base64 20 | docker secret create wp_db_password -
$ docker network create -d overlay traefik
$ docker network create -d overlay mariadb
And deploy stack:
$ docker stack deploy -c docker-stack.yml wp
$ docker stack ps wp
ID NAME IMAGE NODE DESIRED STATE CURRENT STATE ERROR PORTS
aruuw4nc5aog wp_traefik.1 traefik:1.2 ip-172-31-17-95.ec2.internal Running Running 16 minutes ago
lwhjrhsq011x wp_wp.1 wordpress:latest ip-172-31-28-121.ec2.internal Running Running 16 minutes ago
jrddy5uksk3b wp_mariadb.1 mariadb:latest ip-172-31-2-231.ec2.internal Running Running 16 minutes ago
Now we can open our WordPress url https://wp.maddevs.io/ with http to https redirect.
And now we can scale traefik and wp manually:
$ docker service update --replicas 3 wp_traefik
$ docker service update --replicas 2 wp_wp
$ docker stack ps wp
ID NAME IMAGE NODE DESIRED STATE CURRENT STATE ERROR PORTS
aruuw4nc5aog wp_traefik.1 traefik:1.2 ip-172-31-17-95.ec2.internal Running Running 16 minutes ago
lwhjrhsq011x wp_wp.1 wordpress:latest ip-172-31-28-121.ec2.internal Running Running 16 minutes ago
jrddy5uksk3b wp_mariadb.1 mariadb:latest ip-172-31-2-231.ec2.internal Running Running 16 hours ago
r3t531anb8f0 wp_traefik.2 traefik:1.2 ip-172-31-36-181.ec2.internal Running Running 15 minutes ago
0gnadxbm32gt wp_traefik.3 traefik:1.2 ip-172-31-2-231.ec2.internal Running Running 15 minutes ago
7yp7b6igimth wp_wp.2 wordpress:latest ip-172-31-0-161.ec2.internal Running Running 15 hours ago
Or by editing deploy section in our yml and redeploy stack.
We didn't create more complicated infrastructure using another Amazon services like RDS, S3 + CloudFront, Elastic Cache etc but example above shows us the perfect symbiosis of cloud hosting and swarm mode with new docker >v1.13 features. Docker Cloud swarm mode removes our concerns about infrastructure and swarm management.
And it's only beta.