Related, we have this in logging.c:

My first thought here is to start by converting pcmk__remote_ready to be an async function that checks the remote to see if it's ready once and sends a result up to the mainloop. However, one caller of this function is lrmd_poll, which we do not use anywhere but I assume is public API. Do we need to continue keeping it (and lrmd_dispatch) around? If so, we might need sync and async versions of this function.

I think those are the only three tests in cts-scheduler that also have a third argument to SchedulerTest. This is probably something I broke in one of the refactoring rounds.

It's odd that they would only fail when run individually. One clue as to what's going wrong might be that those are the only three tests with the string "Setting effective cluster time" in their summary files, and that string comes from set_effective_date.

Note that cts-scheduler does run with -S in at least two places: