Skip to content

Latest commit

 

History

History
122 lines (85 loc) · 5.38 KB

README.md

File metadata and controls

122 lines (85 loc) · 5.38 KB
Raw

Docker Waiter Build Status

This is an image collection of "waiters", images that are meant to be run as one-off processes, exiting succesfully when the matching service they are waiting for is ready.

A common problem in docker, is expecting that once a cointainer is started, it is also inmmediately ready, so it can interoperate with other services normally, but that is not true for many common services. Data races and faulty deployments can occur spuriously if this is not taken seriously into consideration when designing integration routines. Thus, a sane solution for avoiding those problems would be pinging the service until a pong is returned, indicating that the container is finally ready to operate as expected. Unfortunatelly, the ping interface needs to be implemented differently for every kind of service, and because this task can be rather cumbersome, you might consider using some waiters already implementing those solutions in this repo.

All waiter images will use exactly the same waiter script as default command that can be configured with the following environment variables:

  • WAITER_TIMEOUT (optional, default value is 50): Time in seconds that you are willing to wait to your service. If there is no succesful ping before the timeout, it means that the service is expected to never be ready for the current context, so the process will exit with error code 1.
  • WAITER_ATTEMPT_SLEEPTIME (optional, default value is 5): Amount of time the waiter will wait in seconds after an unsuccesful ping attempt. After that time, a new attempt will begin.
  • WAITER_DEBUG (optional, default: false): If true, it will show in console the response of the ping command executed on attemps, otherwise it will be hidden.

And in order to provide wait logic for a heterogeneous number of services, every waiter here provides a different definition of ping.sh that is useful for its correspoding service. In case you wanna check them, those are located at: services/*/ping.sh

List of Images

  • theypsilon/waiter-mysql:0.1.23: Waits until MySQL is able to execute queries.
  • theypsilon/waiter-rest:0.1.23: Waits until the expected HTTP STATUS CODE is returned from a given HTTP URL
  • theypsilon/waiter-elasticsearch:0.1.23: Waits until Elastic Search server is up and the cluster health status is green or yellow.

How to use the Images

theypsilon/waiter-elasticsearch:0.1.23

Configuration of the Elastic Search Waiter can be done through the following environment variables:

  • ELASTICSEARCH_HOST (mandatory)
  • ELASTICSEARCH_PORT (mandatory)

Example:

$ docker run --name elasticsearch_server -d elasticsearch:1.5.2

51321a23d283005344502b330b5249248f70f0c5ac4cee36fb6356a17f078034

$ docker run -it --rm --link elasticsearch_server:elasticsearch_server \
	-e ELASTICSEARCH_HOST=elasticsearch_server \
	-e ELASTICSEARCH_PORT=9200 \
	-e WAITER_DEBUG=false \
	-e WAITER_TIMEOUT=20 \
	-e WAITER_ATTEMPT_SLEEPTIME=1 \
	theypsilon/waiter-elasticsearch:0.1.23

[WAITER] WAITER_TIMEOUT: 20 | WAITER_ATTEMPT_SLEEPTIME: 1
[WAITER] ....... OK

theypsilon/waiter-mysql:0.1.23

Configuration of the MySQL Waiter can be done through the following environment variables:

  • MYSQL_HOST (mandatory)
  • MYSQL_PORT (mandatory)
  • MYSQL_USER (mandatory)
  • MYSQL_PASSWORD (mandatory)
  • MYSQL_QUERY (optional, default value is select 1): As long as the query produce an Error, the ping will be valid.

Example:

$ docker run --name mysql_server -d \
	-e MYSQL_ROOT_PASSWORD=root \
	-e MYSQL_USER=test \
	-e MYSQL_PASSWORD=test \
	-e MYSQL_DATABASE=test \
	mysql:5.6.28

c03b737671099e1454e65e03cb1d4d1c05fdb4ddfd7bf725112d495ad637605a

$ docker run -it --rm --link mysql_server:mysql_server \
	-e MYSQL_HOST=mysql_server \
	-e MYSQL_PORT=3306 \
	-e MYSQL_USER=test \
	-e MYSQL_PASSWORD=test \
	-e WAITER_DEBUG=true \
	-e WAITER_TIMEOUT=20 \
	-e WAITER_ATTEMPT_SLEEPTIME=1 \
	theypsilon/waiter-mysql:0.1.23 && echo DONE!

[WAITER] WAITER_TIMEOUT: 20 | WAITER_ATTEMPT_SLEEPTIME: 1
[WAITER] ...... OK
DONE!

theypsilon/waiter-rest:0.1.23

Configuration of the REST Waiter can be done through the following environment variables:

  • HTTP_URL (mandatory): URL of the shape http://%SERVICE_DOMAIN:%SERVICE_PORT/route/ping
  • HTTP_STATUS_CODE (mandatory): The expected HTTP STATUS CODE meaning that the resource behind the given HTTP URL is OK, typically 200.

Example:

$ docker run --name elasticsearch_server -d elasticsearch:1.5.2

f14ac8441f2614748e461acd4ab0417d77a59a06e25fab240a88406736d15107

$ docker run -it --rm --link elasticsearch_server:elasticsearch_server \
	-e HTTP_URL=http://elasticsearch_server:9200/_cluster/health \
	-e HTTP_STATUS_CODE=200 \
	-e WAITER_DEBUG=false \
	-e WAITER_TIMEOUT=20 \
	-e WAITER_ATTEMPT_SLEEPTIME=1 \
	theypsilon/waiter-rest:0.1.23 && echo DONE!

[WAITER] WAITER_TIMEOUT: 20 | WAITER_ATTEMPT_SLEEPTIME: 1
[WAITER] .. OK
DONE!

NOTE: Notice that checking elasticsearch by looking the http status code is not sufficient, and you should use the image 'theypsilon/waiter-elasticsearch:0.1.23' instead to know when the elasticsearch clusters are ready.

Contributing

This repository accepts pull requests.

More details at CONTRIBUTING.md

License

LICENSE.md