Even if you do not plan to use Docker to distribute your application you can use it to speedup setup of development environment, for new developers and for new machines. I have a project where we use MongoDb and ElasticSearch, mongodb should be authenticated and ElasticSearch needs to have some special plugin installed.
Time to setup a new machine sometimes is high due to dependencies.
I’m aware that for experienced user setting up mongodb and ElasticSearch is not a complex task, but nevertheless you usually can have some problem.
- Which is the correct version of ElasticSearch/Mongodb that our software is using?
- Are there some Es plugin to install? Which version?
- Which version of JAVA should I install?
- How can I configure a user in Mongodb?
Clearly I wish for a developer not to be worried about these things, thus you can simply use docker-compose to setup most of the requirements. Here is an example.
This is a really simple file, it runs a mongodb image and a specific elasticsearch image. It uses environment variables to let each developer to specify location of data files and password for mongodb. These environment variables: ES_DATA_LOCATION, MONGODB_DATA_LOCATION, MONGODB_ADMIN_PASSWORD could be simple environment variables or you can specify in a .env file contained in the very same directory of docker-compose file.
MONGODB_DATA_LOCATION=g:\Nosql\dockermongo ES_DATA_LOCATION=g:\Nosql\dockeres MONGODB_ADMIN_PASSWORD=123456##
Thanks to these few lines of code I can now start up a mongodb and an elasticsearch instance in mere seconds thanks to the command
docker-compose -f filename.yml up -d
Now the only thing I need to do is installing plugins inside ElasticSearch container, with simple docker-compose exec commands.
docker-compose -f .\filename.yml exec es /usr/share/elasticsearch/bin/plugin install royrusso/elasticsearch-HQ/v2.0.3 docker-compose -f .\filename.yml exec es /usr/share/elasticsearch/bin/plugin install delete-by-query
To simplify everything I create a simple Windows batch file called FirstExecution.bat that call docker-compose up, wait for 20 seconds to give elasticsearch cluster time to start and the install plugin with docker-compose.
You could obtain the very same result modifying the startup script of the container, but in my opinion: using standard elasticsearch container without modification, then installing plugin with standard command line is more stable and simple to understand for people not used to docker and docker-compose.
Now a simple click and the developer can start developing not worrying about database engines.