Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 22 Aug 2004 20:32:05 +1000
From:      Peter Grehan <grehan@freebsd.org>
To:        Andre Oppermann <andre@freebsd.org>
Cc:        cvs-all@freebsd.org
Subject:   Re: cvs commit: src/sys/net rtsock.c
Message-ID:  <41287625.1020800@freebsd.org>
In-Reply-To: <4128673E.F68B68EE@freebsd.org>
References:  <200408212120.i7LLK6YQ092852@repoman.freebsd.org> <4128673E.F68B68EE@freebsd.org>

next in thread | previous in thread | raw e-mail | index | archive | help
> A rtmessage should never ever be dropped.  That would wedge the synchronized
> state of any userland routing daemons.

  That only works with infinite memory or if you can flow-control
the sources of the messages. A drop is easy to cause e.g. by
downing a SONET link that might have many layers of serial ifnets
muxed on top of it. Gated used to handle this by looking for a gap
in sequence number and then falling back to a sysctl to re-read the
routing table/interface list. Hopefully not frequent events !

  Ask the Juniper guys how they handle this case :-)

later,

Peter.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?41287625.1020800>