On 21 January 2016 at 13:53, Lance Richardson <lrich...@redhat.com> wrote: > ----- Original Message ----- >> >> > On Jan 21, 2016, at 1:19 PM, Joe Stringer <j...@ovn.org> wrote: >> > >> > On 21 January 2016 at 11:57, Lance Richardson <lrich...@redhat.com> wrote: >> >> The "conntrack - ICMP related with NAT" test case currently fails >> >> on systems using the nmap version of nc because this version >> >> does not support the -q command-line option. >> >> >> >> Fix this by detecting when the nmap version is in use and using the >> >> corresponding "--send-only" command-line option: >> >> >> >> --send-only Only send data, ignoring received; quit on EOF >> >> >> >> Signed-off-by: Lance Richardson <lrich...@redhat.com> >> > >> > Jarno / Daniele, do you have any thoughts on this approach? >> > >> >> IMO it would be good to have netcat usable for testing. So I’m in favor of >> Lance’s patch. >> >> Jarno >> >> > The base problem is that 'nc' has significantly different >> > implementations if you're on RHEL, Debian or Ubuntu. In commit >> > dc55e9465511dee6c12dbf0edb4ce2d9af57cb15, I avoided this issue by >> > replacing the netcat command with openflow packet-outs, but I think >> > that the approach which Lance is proposing is more elegant. Perhaps we >> > should revert the above commit and apply a similar approach to that >> > test as well. >> >> > Should have mentioned that I've tested via "make check-kernel" on RHEL7 and > Debian 8.2, > with current net-next kernel and Jarno's NAT support patches applied. > > It seems current versions of the openbsd and "traditional" > flavors of netcat support the -q <seconds> option and the nmap flavor appears > to be the only one with --send-only. > > I'd be happy to submit a v2 that also reworks commit > dc55e9465511dee6c12dbf0edb4ce2d9af57cb15.
Thanks, that sounds like a good plan. Please do. _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev