Phil Dibowitz wrote:
>
> Matt Sykes wrote:
>
> > I'm currently blocking all unsolicited incoming icmp. I just learned this
> > is a bad thing, in particular blocking icmp-type 3 code 4.
>...
> > I might be okay, though, because the ipfilter howto appears to say that
> > the appropriate icmp types are allowed for stateful connections. It is
> > not altogether clear to me though.
>
> This is correct.
>
> > In particular, does
> >
> > block in quick on xl1 from any to any
> > pass out quick on xl1 proto tcp from any to any flags S keep state keep frags
> > pass out quick on xl1 proto udp from any to any keep state
> > pass out quick on xl1 proto icmp from any to any keep state
> >
> > allow in icmp-type 3 code 4 for a stateful tcp connection?
>
> Yes.
Are you sure about that? When we started using IP Filter recently (V3.4.25, on
Solaris 2), I thought the same but rapidly found that it wasn't really true,
with a proportion of such ICMPs being logged as blocked. I assumed that was
because the ICMPs were from intermediate "infrastructure" (routers, terminal
server sending source quench for a saturated dialup connection, etc.) rather
than the end-point recorded by "keep state" - the log entries identified the
sources and that explanation certainly seemed to fit. Does that sound
reasonable? Or should the ICMP still have been recognised, so that the
blocking implies something else was wrong?
Our configuration files are currently allowing echo, squench, redir,
unreach, and timex in explicitly, plus any other ICMP that's recognisably
associated with stateful TCP connections.
John Line