In this post I’m going to explain how to create and use a docker image, for easy access to elasticsearch for development purposes. If you don’t know docker, just check it out. In my opinion is the way that everybody is going to distribuite webapps soon.
In the same way that the project has library dependencies, your app usually depends on more things. Let’s say the database or in this case the search engine. Why not distribute the dependency with the code? Well even if you have a VM could be to much.
With containers that is over, now you can distribute with your source a Dockerfile, or even have yor own repository of docker images.
For my dependency elasticsearch a need a new Dockerfile, so the next developer could reproduce my environment very easy.
So this is the content of my Dockerfile.
https://gist.github.com/dahernan/6149626
Now I can build the image with:
$ docker build -t="elasticsearch" .
And after a while if I do:
$ docker images
REPOSITORY TAG ID CREATED SIZE
ubuntu 12.04 8dbd9e392a96 3 months ago 131.5 MB (virtual 131.5 MB)
ubuntu 12.10 b750fe79269d 4 months ago 24.65 kB (virtual 180.1 MB)
ubuntu latest 8dbd9e392a96 3 months ago 131.5 MB (virtual 131.5 MB)
ubuntu precise 8dbd9e392a96 3 months ago 131.5 MB (virtual 131.5 MB)
ubuntu quantal b750fe79269d 4 months ago 24.65 kB (virtual 180.1 MB)
elasticsearch latest ef2487bb289d 52 seconds ago 12.29 kB (virtual 558.7 MB)
I can see my new image elasticsearch there, and finally I can run elasticsearch in that container.
$ docker run -d elasticsearch
$ docker ps
ID IMAGE COMMAND CREATED STATUS PORTS
29fd16250848 elasticsearch:latest /bin/sh -c elasticse 32 seconds ago Up 32 seconds 9200->9200
Now my elasticsearch instance is running in its own container, forwarding the port 9200. So if I want to try to run a elasticsearch cluster, I’m only have to run more containers.