Running the same Shadow experiment twice results in non-deterministic behavior; so far I have observed the following:

• hosts may get assigned different IP addresses
• tgen uses getpid and logs Shadow's PID instead of the virtual process PID
• packet events are executed out of order
• gettid using syscall(SYS_gettid, ...) should be implemented
• futexes are woken in a non-deterministic order
• descriptors are notified in a non-deterministic order
• non-deterministic behavior when retransmitting packets - no evidence in Deterministically iterate status listeners #2109