I assume this depends more on which kinds of tests are being run, but it's relatively common for a node crashing to result in the test coming to a halt and timing out instead of failing.

Major benefits:

• Less time spent running tests. A test that takes seconds/minutes can crash and instead timeout after hours.
• Large logs as a result of timeouts can be quite the challenge to inspect. We have improvements coming through for that, but it seems worthwhile to avoid recording GBs of log data when typically everything one needs to know is in the first ~100MB.

Apologies if this ticket already exists – I felt I had seen one before. I was struggling to come up with the right keywords to find it.