This playbook is a basic collection of settings for running a two node cluster with some basic failover mechanisms on DigitalOcean (referral). The basic idea is to run two CouchDB databases in Master<->Master replication and two Solr Nodes that would listen to the local CouchDB copy.
GlusterFS is used for shared storage, and Thumbor creates thumbnails. Internal Communication between nodes is done via private networking within digital ocean using tinc.
Both nodes are using a DigitalOcean Floating IP that is assinged to one of them. Corosync/Pacemaker is used to automatically switch them in case of failure on the master. Nginx can be used to route the traffic to both nodes.
This is a stripped down version of my in-production so YMMV (a lot actually). Take everything with a grain of salt!
This requires Ansible 2.x. Also, in real life you want to encrypt the
group_vars/* and host_vars/* using ansible-vault.
The WTFPL.