We already have evergreen tasks to run the tiered hook, the timestamp hook and both hooks in combination.   These were intentionally not made part of the PR tests because (in the case of the tiered hook) it runs most of the Python test suite, and that would increase the time of the PR test enormously.  There are ways around that, but they add some complexity.  The downside is that a the test suite has (currently) small coverage of tiered storage, and a change that breaks tiered storage may not be caught in the PR (though it will be caught after merge).

It would seem we need a little more coverage, so we propose running each hook (or combination) and enable a random sample of the tests (with run.py --random-sample N ).  Maybe N=20 will keep these tasks from being the long pole.