Redis Swarm Configuration documented

[trkohler]

should resolve: #2043

my notes & understanding on how it works based on what I read https://github.com/zalando/skipper/pull/2009/files

kubernetes-redis-service-namespace - sets namespace for redis to be used to lookup endpoints
kubernetes-redis-service-name - sets name for redis to be used to lookup endpoints. The only required parameter to redis swarm to be working.
kubernetes-redis-service-port - sets port for redis to be used to lookup endpoints, default value is 6379

skipper is getting all endpoints using kubernetes service for redis cluster
then it updates the list of those addresses each 10 seconds
updates works in background with the separate goroutine
this enables autodiscovery for new redis instances

question: do we need an image / system design of how auto-discovery works? for me it looks like it's simple to understand, but I'm open for suggestions.

[szuecs]

@trkohler if you create an image it's of course great for the reader.

Maybe also tell something about the routesrv case, because if you have a non-kubernetes skipper data-plane and use routesrv, you need to set these options to routesrv and set -swarm-redis-remote to the skipper data-plane to fetch the redis endpoints from routesrv API.

[MustafaSaber]

https://opensource.zalando.com/skipper/tutorials/ratelimit/#redis-based-cluster-ratelimits

[szuecs]

👍

[AlexanderYastrebov]

👍

[MustafaSaber]

👍