John mentioned that there was a time a bunch of tests started failing together and they have similar stack trace. It would make sense to group them together to dedup.

Two ideas were brought up to identify them:

1. Parse the log (stack trace) to understand the root cause.
2. Inspect the occurrence history: if two tests started being flaky around the same time and always fail together, then it's very like that they have the same root cause.
 

John mentioned that there was a time a bunch of tests started failing together and they have similar stack trace. It would make sense to group them together to dedup.

Two ideas were brought up to identify them:

1. Parse the log (stack trace) to understand the root cause.
2. Inspect the occurrence history: if two tests started being flaky around the same time and always fail together, then it's very like that they have the same root cause.

A similar problem in Flake Analyzer:  crbug.com/850319 . Solutions can potentially be shared between Flake Detection and Flake Analyzer.