intermittent rsh 'connection refused'

[Lucas Bergman]

Hi --

I have a repository that is accessed via rsh on my company's local
net, which we've been using for about 9 months.  Once in a while,
though, access is suddenly cut off.  One does 'cvs update', and after
no output for around thirty seconds, one finally gets error messages:

  server.foo.com: Connection refused
  cvs [update aborted]: end of file from server \
    (consult above messages if any)

Both client and server are a recent GNU/Linux, and run CVS 1.11.  My
$CVSROOT is ':ext:server.foo.com:/usr/cvsroot', and $CVS_RSH is not
set at all.  Simply doing 'rsh server.foo.com' from the same client
works like it should.  Whenever this happens, all rsh clients are cut
off simultaneously, not just one or a few.

Normally the problem corrects itself within five to as long as sixty
minutes.  It seems to happen more often when lots of big check-ins are
happening, and the problem seems to correct itself more slowly if a
lot of developers are banging on it, but that's pure anecdote, no real
evidence to back it up.

Lucas