-      # TLDR; synapse has no HA support ;-; (yet)

-      # https://matrix.org/blog/2020/11/03/how-we-fixed-synapse-s-scalability/

-      # Note: we don't have nearly as many users as Matrix.org, so

-      # fine-tuning the values for our use-case would be best

-      # workers:

-      # https://docs.element.io/latest/element-server-suite-classic/advanced-configuration/synapse-section-workers/#worker-types

-      #   client-reader:

-      #     enabled: true

-      #     replicas: 2

-      #   -> We could probably just enable everything, stress test bottenecks and

-      #      increase replication for the bottleneck processes

-      # As far as we are concerned, our deployment is (very) small, sub 70 users

-      # as of right now, so configuring workers is pointless

-      # Synapse will use redis by default for pub/sub IPC when using workers,

-      # so we might as well enable it for caching

-      # redis:

-      #   enabled: true

-      # Also note: workers are more of a scalability thing, not a HA thing

-      # We can't really have a "hot-spare/hot-replication" synapse server either,

-      # because of the way synapse streams deal with caching

-      # (matrix.org itself goes down if the server crashes, so we kinda gotta accept it)

-      # Best thing we can do for HA is making it come back faster, like our GitLab instance

-      # Another option is having a "cold-spare", but I have no idea how to do that.