Details
-
Bug
-
Resolution: Fixed
-
Major
-
6.0.0
-
Untriaged
-
Centos 64-bit
-
Unknown
Description
It's a known limitation that if eventing process is killed externally, it may lose timers that have been dequeued but not run to completion. This is because eventing-producer does not follow a Request & Acknowledge protocol when queuing events to v8 workers. As a result, forcibly killing the eventing producer OS process will cause the timers that have been dequeued from storage but not yet fired by v8 workers to be lost.
Attachments
Issue Links
- blocks
-
MB-31387 timers may be lost in case of producer/consumer crash
- Closed
For Gerrit Dashboard: MB-30772 | ||||||
---|---|---|---|---|---|---|
# | Subject | Branch | Project | Status | CR | V |
99139,4 | MB-30772: Allow deleting timer from outside Go | unstable | eventing | Status: MERGED | +2 | +1 |
109094,43 | MB-30772: Ensure no timer is lost when v8 eventing process is killed | unstable | eventing | Status: MERGED | +2 | +1 |
114537,3 | MB-30772: Ensure no timer is lost when v8 eventing process is killed | unstable | eventing-ee | Status: MERGED | +2 | +1 |