We keep getting Jenkins errors with Broker tests like the one below.
Does Anybody has an idea how to make these work more reliably?
Robin
Do you recall when pending triggers get evaluated? I think they were tied to event processing and if that fully drains and stalls out, then the “when” statements in the test are maybe waiting until the timeout timer happens?
If that’s the situation, maybe add a recurring/scheduled event is enough of a trick to make the test more reliable. Or running on a longer pcap (I think that was the trick I initially went with because there’s no real reason it needs to use wikipedia.trace like it currently does).
Or maybe an overhaul of the main event and I/O loop would help 
- Jon
Do you recall when pending triggers get evaluated? I think they were
tied to event processing and if that fully drains and stalls out, then
the “when” statements in the test are maybe waiting until the timeout
timer happens?
Yes, it does take place at the time of event draining, but that should
keep happening even without new events being queued, as the draining
executes regularly in any case.
Or maybe an overhaul of the main event and I/O loop would help
Robin