Yannick Pereira-Reis bio photo

Yannick Pereira-Reis

DevOps (docker swarm, haproxy, CI/CD, ELK, prometheus, grafana, ansible, automation, RabbitMQ, LVM, MySQL replication...) and fullstack web developer Symfony 2/3/4/5 + VueJs in Valence (France).

Twitter LinkedIn Github

As a big fan of elasticsearch, I very often use this extraordinary tool to make POCs, test applications like logstash and Kibana, or more generally, to discover the lastest possibilities allowed by elasticsearch (percolator, aggregations, sharding, scalability,…).

But, it’s often very complicated to set up a full elasticsearch cluster for a POC, a simple test or a dev environment.


On the other side, it’s very easy to use the minimalist configuration set up by default when using elasticsearch, but it’s quickly limited when we want to deal with problematics like:

  • Sharding on many physical nodes (routing, filters, rack, …).
  • Network cluster configuration (unicast, multicast, ip:port, …)
  • Snapshot/recovery configuration
  • Specific configurations (load balancer, data nodes et master nodes)
  • Network failure simulation

So I decided to create a project allowing everybody who wants to experiment these kind of simulations, to do so:


This project gives you the ability to start a cluster in seconds with a single simple bash command :

vagrant up

You should see something like:

$ vagrant up
Cluster size: 5
Cluster IP:
Bringing machine 'vm1' up with 'virtualbox' provider...
Bringing machine 'vm2' up with 'virtualbox' provider...

How to start the cluster

Simply clone this repo: https://github.com/ypereirareis/vagrant-elasticsearch-cluster

And follow README.md file instructions.