Closed mcollina closed 6 years ago
We run a 62 hours test with aedes deployed in a container. The issue is greatly reduced: over 56 millions of messages only 4 messages have been leaked. Thus there is still an unknown condition that make it happen, but the module is way better with this PR.
We had previously run several smaller test with the same result in which a tiny minority of messages have been leaked.
If you can create a repro for that condition, I would be extremely happy to do a fix for that. Tracking down a 6.15384615e-8 probability of leakage is extremely hard.
Yes I know, we are struggling creating an example that maximizes the number of occurrences to a debuggable number... I'll let you know, thanks
Fixes https://github.com/mcollina/mqemitter-mongodb/issues/9
@teomurgi can you please check if this fixes it?