Resolve "KafkaReporter throws 'reported timestamp greater than x' in streaming jobs"
requested to merge 53-kafkareporter-throws-reported-timestamp-greater-than-x-in-streaming-jobs into master
This merge request does 4 things:
- Takes care of #53 (closed).
- Fixes some typos with grabbing the nicknames of classifiers for topic naming within various stream executables.
-
stream.evaluate
: Actually check whether a classifier has a trained model before doing evaluation. - Ignores error handling for
KafkaReporter.retrieve()
. Most of the time, the error is just caused by an EOF since there's no new buffers in the queue, which we can safely ignore. I'll reimplement this in here and inKafkaClassifierData
and filter by certain messages in a more robust way when we have a stable pipeline up and running and have a better idea of what errors we can and shouldn't ignore.
Closes #53 (closed)
Edited by Patrick Godwin