On Saturday Jul 28th at around 15:48 UTC all the eventlogging processors stopped working all of a sudden. From the logs it seems that this happens (for each processor):
- the processor's consumer (via kafka-python) gets stuck reading one or more events
- the processor's producer sockets are torn down by the Kafka brokers after some minutes (corresponding to the default idle timeout before cleaning up "unused" sockets).
The next step is try to figure out what causes this mess and let Eventlogging restart.