While Parsoid logs performance metrics during roundtrip testing, we don't have anything for the production cluster itself.
It would be useful to add instrumentation to different pieces of the parsoid pipeline for both wt2html and html2wt directions.
Once Parsoid is integrated with RESTbase, aggregate full-request numbers will also be logged there, but, finer-grained numbers for different stages would nevertheless be useful.
Completing this task will require figuring out some of the following:
- How does this fit in with the editing performance metrics?
- What pieces of the Parsoid pipelines need instrumentation?
- Where to send perf events to?
- What format / schema is needed by the perf event sink?
- What node library / binding to use to help with the instrumentation?
- What front-end is currently being used for performance visualization?
The Parsoid team can help with some of these questions.