When I briefly checked out WebPageReplay tests in T360897 I could see that our direct tests is much more unstable than direct tests against the same URLs that I run on an another server. What's going on there? Let me first collect metrics and differences.
Description
Description
Related Objects
Related Objects
- Mentioned Here
- T360897: Investigate Chrome 123 alerts with WebPageReplay
Event Timeline
Comment Actions
I compared the data between https://grafana.wikimedia.org and the one I have on dasboard.sitespeed.io and the difference is quite small, I cannot say that our metrics are broken. Let me compare configurations, maybe there are some tuning that can be done to make it a little better.
Comment Actions
The difference is that on our instance have "traceCategory": ["disabled-by-default-v8.cpu_profiler"] that gives us traces from our code so I think we should continue to do that.