Peer pods autodiscovery within kubernetes cluster

Hi,

I am experimenting with a simple standalone package to automatically keep pod peers up-to-date within a kubernetes cluster:

https://github.com/udhos/kubegroup

Recipe currently looks like this:

groupcachePort := ":5000"

// 1. get my groupcache URL
myURL, errURL = kubegroup.FindMyURL(groupcachePort)

// 2. spawn groupcache peering server
pool := groupcache.NewHTTPPool(myURL)
server := &http.Server{Addr: groupcachePort, Handler: pool}
go func() {
    log.Printf("groupcache server: listening on %s", groupcachePort)
    err := server.ListenAndServe()
    log.Printf("groupcache server: exited: %v", err)
}()

// 3. spawn peering autodiscovery
go kubegroup.UpdatePeers(pool, groupcachePort)

// 4. create groupcache groups, etc: groupOne := groupcache.NewGroup()

Does it look reasonable!?

Start by reviewing the linked kubegroup package and the issue's usage flow through kubegroup.FindMyURL, groupcache.NewHTTPPool, and kubegroup.UpdatePeers. Determine whether Kubernetes peer autodiscovery belongs in groupcache and what behavior or integration would be required; the issue currently asks only whether the approach is reasonable.