Currently, when using MongoDB as your backend and having more than one machine for Janus you need a Redis instance to make sure Janus propagates the changes. This is a bad implementation and we should probably change it to:
Using the gossip protocol for clustering
On a change, in the API we should broadcast the change
Once a message is received we should reload the mutex (like it's done today)
Currently, when using
MongoDB
as your backend and having more than one machine for Janus you need aRedis
instance to make sure Janus propagates the changes. This is a bad implementation and we should probably change it to: