I’ve recently been seeing some stability issues with Redis that I have running for the k8s.social Mastodon instances. After looking into it I realised that I had it configured in a master/replica architecture but I wasn’t actually making any use of the replicas as Mastodon was configured to do everything via the master. There’s two things wrong with this - firstly I’m wasting resources by having the replicas running but more importantly I created a single point of failure. When the Re...
