> When my yp client (NetBSD) looses his yp server (a sun, or anoter NetBSD
> running an old version of FreeBSD's ypserver) for some reason, it start
> sending a lot of broadcast messages (several in a second) which make
> portmap and ypserv eating all the CPU on the server, in such a way that
> they don't respond to others yp queries. The NetBSD client also don't
> recover from this kind of failure, the only thing to do is to restart
> ypbind on the client.

I have noticed the same behaviour, but on NetBSD/Alpha (with Digital
UNIX YP servers).  I never reported it as a bug because i was never
able to replicate it 'on demand' and figured that it was an
alpha-specific bug...  "i guess not."

Worth noting: in my case, it doesn't send "several in a second," it
appears to be sending "as many as it can, as quickly as possible."
A flood of broadcasts, if you will.

At various times, this has caused every single one of my YP-using
machines here at CMU to be removed from the network by the network
admins...  very annoying, especially since they sometimes don't send
mail (or often, don't send the right person mail) informing about
it...


cgd