I'd have to agree with Pekka. We've just gone through this with ATIS
and "reliable" accounting, which is quite a bit messier than this
problem. So long as the log indicates that something bad has happened,
it seems to me that you're covered. Of course, with reliability comes
some amount of complexity. If an implementation detects the problem,
say, through the other side not ACKing data, it's probably a good idea
to switch to one's backup log server, right? I know of no syslogger
that behaves in this manner today, although perhaps there are some out
there. That doesn't mean we should not provide the function.
Finally, it seems to me we are all arguing over non-normative text.
Choice of level of reliability is generally a deployment concern.
Eliot
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf