Hi !
I'm working on adding a reverse proxy using Hipache (
https://github.com/dotcloud/hipache ) to openruko in order to complete the stack and provide high availability. The only missing features of hipache are :
- reporting idle instances
- redirect to configuration 40x / 50x error pages if setted
All we need to do is add redis to openruko, and report app domains, etc. into redis rather than in postgre (or in both of them ?) we can then listen to dead instances thanks to redis pubsub, then it should work like a charm.
What do you think ? How should we handle hipache modification ? a maintained openruko forked ? try to put what is necessary into hipache ? just add request events in redis pubsub, and add a listener program to emit idle events ? add an "in redis" configuration for 40x/50x pages ?
Cheers,
Florian