Date: Wed, 06 Sep 2006 09:53:09 -0700 From: Sam Leffler <sam@errno.com> To: Andre Oppermann <andre@freebsd.org> Cc: "Eric W. Bates" <ericx_lists@vineyard.net>, freebsd-net@freebsd.org Subject: Re: showing esp tunnels in routing table Message-ID: <44FEFCF5.2010409@errno.com> In-Reply-To: <44FEF4B4.3000807@freebsd.org> References: <44FEDD18.8060506@vineyard.net> <20060906144002.GI30554@catpipe.net> <44FEE301.2090008@vineyard.net> <44FEEFB9.2060408@errno.com> <44FEF4B4.3000807@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Andre Oppermann wrote: > Sam Leffler wrote: >> Eric W. Bates wrote: >>> Phil Regnauld wrote: >>>> Eric W. Bates (ericx_lists) writes: >>>>> When you establish an esp tunnel, the subnets on the remote end of the >>>>> tunnel do not seem to appear in either "netstat -nr" or 'route get >>>>> xxx.xxx.xxx.xxx' >>>>> >>>>> Is there a way to display those routes other than using setkey to dump >>>>> the SPD's? >>>> No, because there are no routes. The IPSec layer "hijacks" the >>>> packets >>>> and they are encapsulated before the routing table gets a chance >>>> to see them. >>>> >>>> You would have to setup transport ESP + gif/gre tunnels to see >>>> routing >>>> entries. >>> Apparently, openbsd's implementation of netstat allows one to view ESP >>> 'flows' (I believe that is how they refer to them) by examining the >>> family 'encap' >>> >>> netstat -rnf encap >>> >>> We have no such equivalent? >> >> openbsd integrated the SAD w/ the routing table; something I've wanted >> to do forever. > > Having it in a separate radix tree (aka routing table) is just fine. > Integrating it with the IPv4/6 routing table is evil and would cause > me some heartburn. > The main point is to integrate routing decisions. I've also felt the locking overhead in IPsec could be significantly reduced by flattening the data structures. I don't care how things are implemented. Sam
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?44FEFCF5.2010409>