docker with ansible wait for database - docker

Question

I try to deploy docker with ansible. I have one docker database container, and in other container is my web app, and I try to link this two container. The problem is that database container didn't have a time to configure itself and a web container is already started. My ansible playbook look something like:
...
- name: run mysql in docker container
docker:
image: "mysql:5.5"
name: database
env: "MYSQL_ROOT_PASSWORD=password"
state: running
- name: run application containers
docker:
name: "application"
image: "myapp"
ports:
- "8080:8080"
links:
- "database:db"
state: running
How to determine if database is start? I try with wait_for module, but that didn't work. I don't want to set timeout, it's not good option for me.

Answer 1

wait_for does not work for the MySQL docker container because it only checks that the port is connectable (which is true straight away for the Docker container). However, wait_for does not check that the service inside the container listens the port and sends responses to the client.
This is how I am waiting in the ansible playbook for the MySQL service becoming fully operational inside the Docker container:
- name: Start MySQL container
docker:
name: some-name
image: mysql:latest
state: started
ports:
- "8306:3306" # it's important to expose the port for waiting requests
env:
MYSQL_ROOT_PASSWORD: "{{ mysql_root_password }}"
- template: mode="a+rx,o=rwx" src=telnet.sh.j2 dest=/home/ubuntu/telnet.sh
# wait while MySQL is starting
- action: shell /home/ubuntu/telnet.sh
register: result
until: result.stdout.find("mysql_native_password") != -1
retries: 10
delay: 3
And the telnet.sh.j2 is
#!/bin/bash -e
telnet localhost 8306 || true

Answer 2

To avoid the sh and I don't normally have telnet installed...
- name: Wait for database to be available
shell: docker run --rm --link mysql:mysql mysql sh -c 'mysql -h"$MYSQL_PORT_3306_TCP_ADDR" -P"$MYSQL_PORT_3306_TCP_PORT" -uroot -p{{mysql_password}} || true'
register: result
until: result.stderr.find("Can't connect to MySQL") == -1
retries: 10
delay: 3

Answer 3

As etrubenok said:
wait_for does not work for the MySQL docker container because it only checks that the port is connectable (which is true straight away for the Docker container). However, wait_for does not check that the service inside the container listens the port and sends responses to the client.
Using Andy Shinn's suggestion of FreshPow's answer, you can wait without needing a shell script or telnet:
- name: Wait for mariadb
command: >
docker exec {{ container|quote }}
mysqladmin ping -u{{ superuser|quote }} -p{{ superuser_password|quote }}
register: result
until: not result.rc # or result.rc == 0 if you prefer
retries: 20
delay: 3
This runs mysqladmin ping ... until it succeeds (return code 0). Usually superuser is root. I tested using podman instead of docker but I believe the command is the same regardless. |quote does shell escaping, which according to the Ansible docs should also be done when using command:

Answer 4

This works for me just fine:
- name: get mariadb IP address
command: "docker inspect --format '{''{ .NetworkSettings.IPAddress }''}' mariadb-container"
register: mariadb_ip_address
- name: wait for mariadb to become ready
wait_for:
host: "{{ mariadb_ip_address.stdout }}"
port: 3306
state: started
delay: 5
connect_timeout: 15
timeout: 30

Answer 5

Use wait_for module. I'm no expert on MySQL but I assume there would be some port or existence of file or message in some log file etc. you can check to find out if the DB is up or not.
Here are examples of wait_for copied from the link above.
# wait 300 seconds for port 8000 to become open on the host, don't start checking for 10 seconds
- wait_for: port=8000 delay=10
# wait 300 seconds for port 8000 of any IP to close active connections, don't start checking for 10 seconds
- wait_for: host=0.0.0.0 port=8000 delay=10 state=drained
# wait 300 seconds for port 8000 of any IP to close active connections, ignoring connections for specified hosts
- wait_for: host=0.0.0.0 port=8000 state=drained exclude_hosts=10.2.1.2,10.2.1.3
# wait until the file /tmp/foo is present before continuing
- wait_for: path=/tmp/foo
# wait until the string "completed" is in the file /tmp/foo before continuing
- wait_for: path=/tmp/foo search_regex=completed
# wait until the lock file is removed
- wait_for: path=/var/lock/file.lock state=absent
# wait until the process is finished and pid was destroyed
- wait_for: path=/proc/3466/status state=absent
# wait 300 seconds for port 22 to become open and contain "OpenSSH", don't assume the inventory_hostname is resolvable
# and don't start checking for 10 seconds
- local_action: wait_for port=22 host="{{ ansible_ssh_host | default(inventory_hostname) }}" search_regex=OpenSSH delay=10

Answer 6

I was able to use wait_for like this:
- name: "MySQL - Check mysql - Wait for mysql to be up"
wait_for:
host: 127.0.0.1
port: 3306
search_regex: "(mysql_native_password|caching_sha2_password)"
This way it will wait for the port o be up and for the service to send some data.
The drawback is that the output may change with mysql versions and configurations. In the example are the strings for mysql 5.5 and 8.0. Adjust for your use cases.
An alternative, avoiding running wait_for, command or shell, may be to retry some mysql command several times until it succedes:
- name: "MySQL - Check mysql - if it responds"
mysql_info:
login_user: root
login_password: "{{ mysql_root_password }}"
filter:
- version
register: mysql_result
until: mysql_result is not failed
retries: 5
delay: 10