need some level of alerting should the OOM killer strike

I still get quite often cases where my Kafka broker gets nuked by the OOM killer. We need a way to systematically report this (ochopod will auto-restart the defunct process in any case). Not sure at this point what's the best. We could go with web-hooks..
