Re: [vpp-dev] Question regarding ICMP NAT behavior

2020-05-15 Thread Jon Loeliger via lists.fd.io
On Fri, May 15, 2020 at 2:38 PM Jon Loeliger wrote: > > So, here is the same scenario using vppctl to set up the test case. > > vpp# nat44 add static mapping icmp local 192.168.0.53 external outside > vpp# show nat44 static mappings > NAT44 static mappings: > local 192.168.0.53 external 192.168.

Re: [vpp-dev] Question regarding ICMP NAT behavior

2020-05-15 Thread Jon Loeliger via lists.fd.io
On Thu, May 14, 2020 at 11:23 AM Jon Loeliger via lists.fd.io wrote: > Did the ICMP mapping open more than was expected or intended here? > > I chased this down in the code a bit, but I'm not sure what the _intent_ > is supposed to be. > When "address only" is true (ie, both ports are 0), then th

[vpp-dev] Question regarding ICMP NAT behavior

2020-05-14 Thread Jon Loeliger via lists.fd.io
Hi vpp-devers, We have a report of an unexpected behavior when using a static NAT with ICMP. It appears that configuring an outside interface to allow ICMP also allows forwarding of all protocols as well. If you start with, say, a blocked TCP on port 22 and an SNMP on port 161, then adding a NAT