The test triggers 100 timers and at the end it fails if some timers were not executed on both server and client.

Using a fast machine with plenty of memory I am seeing that even when the test passes as few as 8% of the timers execute on client, for my failed runs I am seeing 100% running on server.

I believe everything is operating normally but for the purpose of this test there is no constraint to back up the assumption the timers will be distributed.