> The fact that that the FDs are duped shouldn't change anything to the
> events reported: it works while the number of FDs is less than
> FD_SETSIZE (epoll_wait() maxevents argument).

That assumes that epoll_wait() is supposed to return *all* ready fds.  But that is not possible because maxevents is finite.  If you want all events then obviously you may need to call epoll_wait() multiple times.

> I just used dup() to make it easier to test, but you'll probably get
> the same thing it your FDs were sockets connected to different
> endpoints.

This is the part I disagree with -- I think it makes all the difference.  Please try making such a modification.

>>         while True:
>>             ready_writers = set(fd for fd, evt in
>>                                 ep.poll(-1, MAXEVENTS) if fd != r)
>>             if ready_writers.issubset(seen_writers):
>>                 break
>>             seen_writers |= ready_writers
>
>Of course it does, since the returned FDs are a subset of all the
>ready file descriptors.
>
>The point is precisely that, when there are more FDs ready than
>maxevents, some FDs will never be reported.

The program can only terminate when the outer

    while all_writers - seen_writers:
        ...

loop terminates.  So seen_writers == all_writers, and every fd has been reported.