Excessive requests to kubernetes apiserver

[rrichardson]

Apologies if this is a known issue, but in a multi-master k8s cluster, it seems that Traefik receives and then responds to every event which comes across the k8s bus, included leader election information, which is very chatty. Then the nearest that I can tell, it responds to every event with an apiserver request. The resulting effect is that my k8s master (master) is pegged at 100% cpu. Is there a way to filter the events to which Traefik responds?

[Hades32]

Even on a single node k8s installation, traefik+kube-apiserver use up ~150% CPU on my 4 code ARMv8 server! When traefik is stopped it sits comfortably below 20%...

[emilevauge]

This should be solved in the next release with #836.
@errm WDYT?

[errm]

Yes this should be fixed by #836, chattiness should be significantly reduced...

[Hades32]

FYI: current master has great performance now. My little ARM board is now actually usable again. Great work!