Re: address inertia

Henry Spencer <[email protected]> Tue, 18 Feb 2003 18:26:16 -0500 (EST)
Newsgroups gmane.network.freeswan.devel
Message-ID <[email protected]>
On Tue, 18 Feb 2003, Jim Carter wrote:
> > In the wonderful world of IPsec, it's not that simple.  How can you send
> > an error reply which the other end can *trust*? ...
> 
> Agreed, east can't reasonably know if the error reply truly came from west.
> How's this for some heuristics that save the situation?

Yes, you can do things like rate limiting which will help greatly (until
your server crashes twice in fast succession... :-().  However, given that
any solution based on error reports rebuilds the connection only after
traffic has tried to get through, and given that rebuilding takes a bit
of time, it really is better to save this as a fallback method.

Address inertia lets you rebuild right away after a crash, rather than
waiting for error reports.  It's less than perfect, so an error-report
strategy is desirable to catch the ones that fall through the cracks,
but the details of that strategy are a lot less crucial if it handles
only exceptional cases.

At one point, when I was still working for FreeS/WAN, we had an Internet
Draft, intended to be an Informational RFC, that addressed a bunch of
these issues.  Unfortunately, *it* seems to have fallen through the cracks
since... 

                                                          Henry Spencer
                                                       [email protected]