Re: [patch] spurious fails with rsh target

[Felix Lee]

Daniel Jacobowitz <address@hidden>:
> This patch means you're going to run failing tests five times...

ah, hm.  that's a difference between rsh and ssh.  I didn't test
with ssh.  rsh doesn't return the exit status of the remote
command, but ssh does.  the fix is to match on specific error
messages and only retry on those.

> FWIW, I've never had the problem you describe when testing with RSH;
> I can't remember exactly how, but you can configure most systems to
> reuse more quickly and in practice it causes no harm.

I couldn't find a way to modify TIME_WAIT duration on linux
without building a new kernel, and it seems to me difficult to
choose a good value, since it affects all tcp connections, not
just the special case of rsh to a fast testing target on the same
LAN.  increasing the privileged port range is also difficult.
retrying seemed like the easiest option.
--