Use monitoring-bus directly for ExecutedActionMetadata metrics
Description
This MR switches the publishing of metrics relating to ExecutedActionMetadata
metrics to have the scheduler do the publishing directly using the singleton monitoring_bus
. This simplifies the logic and allows removing some methods used only for this purpose out of scheduler.py
and server.py
, and is consistent with how we publish other metrics.
In addition, I added two new metrics related to ExecutedActionMetadata
that were missing:
- Total time spend in the worker
- Time Action spend in queue (we only calculated the average previously)
There don't appear to be any existing tests that touch this part of the code, but if there's a good way to test that metrics are published I'd be more than happy to add some tests.
Edited by Jeremiah Bonney