Can't load replay schedule for max_steps failures about shuttle HOT 2 OPEN

Yeah, I've had a similar experience with large failures being very unwieldy and failing to deserialize when replaying, but never had a chance to track it down.

Have you tried configuring the test to persist failures directly to a file, and then replaying from that? Something like this:

let scheduler = shuttle::scheduler::RandomScheduler::new(1000);
let config = shuttle::Config {
    failure_persistence: shuttle::FailurePersistence::File(None),
    ..Default::default()
};
let runner = shuttle::Runner::new(scheduler, config);
runner.run(|| my_test());

That might at least behave a little better.

Your second problem sounds like a potential non-determinism issue, but without being able to reliably replay the original failure it's tricky to be sure.

from shuttle.

I think you're right that the second issue was just non-determinism: I accidentally was using real rand instead of shuttle::rand somewhere else at that commit.

from shuttle.