I noticed over the past few months that Bisect/Pinpoint often simply stops working and does not report that fail state to the bug. Thus, we end up with a lot of stale bugs and a very low confidence in our perf infrastructure e.g. https://bugs.chromium.org/p/chromium/issues/detail?id=893679#c4

In order to double-check for selection bias, as I mostly triage bugs that are over the SLO I did a quick discussion with a few folks working with Telemetry-based benchmarks and they see the same thing: A too low chance of a bisect/pinpoint job to actually work.

Especially the mobile bots seem to be affected by this.

Please fix.