feeder/rethinkdb-cluster repository overview

Checkout the source here.⁠

This image is layer on-top of the standard rethinkdb image to make it easier to setup a cluster on swarm.

It's inspired a lot thanks to the great walkhrough by Stefan Prodan⁠, but using this image you do not need the secondary/primary steps.

It uses the swarm DNS to request all nodes in the rethinkdb-namespace, and passes it on to the rethinkdb startup command.

#!/bin/bash
JOINS=`getent hosts tasks.rethinkdb | awk '{ print $1 }' | sed -e 's/^/--join /' | tr '\n' ' '`
echo "Starting rethinkdb using $JOINS"
rethinkdb --bind all --no-http-admin $JOINS

If there are running rethinkdb nodes, this morphs into:

rethinkdb --bind all --no-http-admin --join 10.0.0.2 --join 10.0.0.2

If no running rethinkdb nodes, it works out well too:

rethinkdb --bind all --no-http-admin

However, if the node restarts, it will connect itself to the other running nodes.