I did have a look at "missing logs"

https://github.com/haproxy/haproxy/blob/master/scripts/run-regtests.sh#L406-L412

that code tries to detect whether test failed or passed.
if test was timed out (and thus killed), it is considered as "passed".

and there's no way to detect "timed out" tests (I could not find easy one).

I would keep all logs all time (there's cleanup in case of success at the end), but I'm not sure why that complicated machinery was initially created.

is keeping all logs (in case of failure) bad idea ?

Originally posted by @chipitsine in #3336