At some point during pyres_worker startup, the Python pseudo-random number is seeded. Since pyres_worker doesn't generate any random numbers during normal operation, all jobs spawned from a single worker share the same PRNG state.
Found because my jobs were generating the same result from random.randint(0, 100000000) hour after hour. I'll submit a patch.
At some point during pyres_worker startup, the Python pseudo-random number is seeded. Since pyres_worker doesn't generate any random numbers during normal operation, all jobs spawned from a single worker share the same PRNG state.
Found because my jobs were generating the same result from
random.randint(0, 100000000)
hour after hour. I'll submit a patch.