On (03/11/10 11:53), Jean Paul Degabriele wrote:
> 
> Hi, Thanks for your reply
> 
> the version is opensolaris 5.11 snv_111b

A lot of changes have gone in since that version, so it might be hard
to nail down the root-cause, but what's the mtu of the tunnel?

Things are a little easier to debug with the latest solaris kernel
because we could insert dtrace probes or breakpoints in 
ip_drop_input and find out why the icmp error is not being sent out..

--Sowmini
> 
> Jean Paul
> 
> On Wed, Mar 10, 2010 at 9:15 PM, <[email protected]> wrote:
> 
> > On (03/10/10 15:49), Jean Paul Degabriele wrote:
> > > Hi,
> > >
> > > I was referred by Dan McDonald to post this problem that I encountered on
> > > this mailing list.
> >    :
> > > interface of H1. IP packets are being sent from H2, fragmented en route
> > and
> > > then tunneled (Hence only the inner datagram is fragmented). Now if at H1
> > > during IP reassembly a fragment is missing H1 should generate and ICMP
> > time
> > > exceeded message, however no ICMP packets are being generated.
> >
> > What OpenSolaris version are you running (what's the output of 'uname -a')?
> >
> > > If the tunnel is removed an ICMP packet is generated so the tunnel seems
> > to
> > > be creating the trouble. From netstat and kstat we can observe that
> > > ipReassmFailed is being incremented but no packets seem to be dropped.
> > >
> > > My question is: Is it meant to be this way, is it a possible
> > configuration
> > > problem, or a bug?
> >
> > it should send an icmp time-exceeded, so it would help to know which snv
> > version is running.
> >
> > --Sowmini
> >
> >
> >
_______________________________________________
networking-discuss mailing list
[email protected]

Reply via email to