Picking a super simple looking jstest: jstests/core/js/jssymbol.js

• Running it in the core suite takes 4 sec
• It’s included in ~100 suites (tasks)
• Those tasks are included in ~60 unique build variants
• This JS test is included in ~1500 variant/task combinations
• Combine this math with longest running tests for another multiplier
• * Eg. txn_commit_optimizations_for_read_only_shards.js averages 10min runtime, included in 96 task/variant combos
• Variants are triggered at different frequency, so I’m not sure yet what’s the runs/day

This test (and others like it) might:

• be needed everywhere just for extra safety (we can likely be smarter)
• be useless and be deleted (difficult to discern)
• be included in too many suites (need domain input to understand test-intent)
• need only be included in a subset of these suites (more intent to know which ones)
• not ever need to be include in any passthrough suites (to live in something like a new core_nopassthrough suite)