Running ElasticSearch locally - its not so bad… right?
For such a powerful distributed system, running ElasticSearch locally is surprisingly doable. Assuming you already installed Java on your machine - and set up the appropriate apt repositories if you are on Debian - a simple [brew|apt-get] install will do the trick. Run elasticsearch with the appropriate configuration from the command line, and
should show your local instance is up and running. If you are only performing simple operations, this local instance will suffice - and you probably don’t need to read the rest of this post. But if you plan on delving deeper into ElasticSearch - testing performance, experimenting with parallelization, and fire drilling disaster recovery - you’ll need more than the single node setup from brew install.
I found myself in said situation earlier this week, wanting to analyze the number of host’s impact on query time. Using Vagrant, a cross-platform program to manage virtual machines, and Ansible, a popular server provisioning and automation platform, I created virtual-elasticsearch-cluster. With virtual-elasticsearch-cluster, setting up a multi-host, multi-node virtual ElasticSearch cluster on your local machine is trivial. Clone the repo from Github and run
and you’re in business. The master nodes runs on the IP address http://192.168.2.4 and listens on the standard port 9200. Running
will show the existence of 2 master nodes and 2 data nodes, spread across three different virtual machines. Stopping the cluster is equally simple - just run vagrant halt from the project directory. If you are done with the cluster and no longer want the virtual machines occupying disk space, run vagrant destroy.
Because of the flexibility of Vagrant and Ansible, you can use virtual-elasticsearch-cluster as a template for any ElasticSearch configuration. Want a different ElasticSearch version? Modify ansible/playbook.yml. Want to access your master through a different private IP address than 192.168.2.4? Edit Vagrantfile. The options go on and on. In the future, I’m considering improving install.sh so you can set configuration options when you run the script, and without having to manually edit files.
I hope you find virtual-elasticsearch-cluster as helpful as I have. As always, please reach out through Github, Twitter, or email ([email protected]) with any contributions, issues, or feedback. Happy hacking.