The developer experience for logging exceptions to the Application log (
MB-31920) is poor.
Right now when in a dev cycle working on a new script it is a bit perplexing when you wait on average 30 seconds (and up to 60 seconds) to see the exception (along with a count) be emitted.
- We could always the very first message, i.e. count===1 immediately the first time we see the exception after any deployment or any resume.
- Then we fall back to the current behaviour of summarization only on count === 2-N ( or minute two+ on 1-N) summarize and emit once per minute.
This would speed up initial debug feedback to the developer for quicker understanding of the issue and still limit overwhelming logging.
|For Gerrit Dashboard: MB-45874|
|152072,1||MB-45874: New exceptions once every logging period by each thread||unstable||eventing||Status: ABANDONED||0||0|
|159192,2||MB-45874: New exceptions once every logging period by each thread||unstable||eventing||Status: MERGED||+2||+1|
|159402,2||MB-45874: New exceptions once every logging period by each thread||cheshire-cat||eventing||Status: MERGED||+2||+1|