The logging pipeline is getting many features we want to reliably work. In order to better diagnose its behavior, deep diagnostic data is needed. Some of this data was historically put into the tags field which would serve us better as a human interface rather than a diagnostic interface.
Must integrate with testingMust be easily enabled/disabled- Can provide deep information about how the pipeline is working
- Which input did <event> come from
- Which filters did <event> hit?
Which output was <event> assigned to?Which bucket was <event> assigned?Which partition (if any) was <event> assigned?