Catégorie : Terminal

When working with MongoDB, it surely comes a time when you want to try to make it scale. First, by adding replica set, then sharding and at least mix it together to create a simple cluster with replication on each shard. Then, you might want to test monitoring or other functionality, so you will need to launch your cluster, configure it.

If everything goes right, you’ll only need to start all your MongoDB instances. Not a big deal, easy to do it at each restart of your computer, just copy paste the right commands. But what if I need a fresh clean cluster to make my test? To answer this question, I wrote a simple script to easily deploy a local mongo cluster composed of one query router, three configuration routers and two shards, each being a replica set composed of one primary mongod and two secondary mongod.

This simple project is on my Github under the name simple-mongo-cluster. The readme describes on which port you’ll find which mongo. The script provides 5 operations: {init|start|configure|stop|clean}. I will quickly describe each of it.

The init operation creates all directories needed to store logs and data. All mongo instances will be started with the start parameter. To configure the cluster, use configure. For the cluster to be configured, you might need to call configure twice, because of shards sometimes not being added at first call. The stop operation is just about shutting down the whole cluster. At last, clean deletes all directories containing the cluster configuration files, data and logs so that you can start a fresh new cluster.

I hope it will help other developers who want to easily test and discover MongoDB cluster possibilities in a local environment as it helps me. If you have any suggestions, just let me know or fork the project, add your script and send me a pull request.