EventStreams is backed by the analytics Kafka cluster, which has its topics mirrored from main Kafka. EventStreams could run on main Kafka (which would give it better multi DC support), but we chose to run it from the larger eqiad only cluster, so users connecting from the open internet would not be able to potentially affect the operation of the main Kafka clusters, which are more critical.
We need to move EventStreams' Kafka from analytics to jumbo. This will mean that offsets will change, which will disrupt connected SSE/EventSource clients. When they reconnect, they will provide invalid offsets to EventStreams. Fortunetly, node-rdkafka will just disregard these invalid offsets and start consuming at the end of the stream. This means that during the switchover, clients may lose a few messages.
(Another complication is that events are currently mirrored from main-eqiad to jumbo-eqiad via MirrorMaker processes running on the analytics Kafka brokers. I'd prefer to move these instances to the jumbo Kafka brokers, but the main Kafkas are 0.9, and the newer MirrorMaker version can't consume from these older clusters. So, we need a place to continue to run the 0.9 MirrorMaker version. I'll leave this on the analytics Kafka brokers for now, until we get all clients off of analytics. Then we can think of what to do about MirrorMaker before we start decomming the analytics Kafka brokers.)
Anyway, for EventStreams:
[] Merge puppet patch, bounce EventStreams.