Sorry one other minor thing I just noticed is that this isn't really a ci fix, it's a test fix. So I think the patch title should have "tests" instead.
On 10/29/26 10:38 AM, Jacob Tanenbaum wrote: > tcpdump treats UDP port 8472 as VXLAN traffic and attempts to print the > inner Ethernet frame when it sees 8472 as the source port number. > > When get_random_port_number() returns 8472 it causes the check > > AT_CHECK([grep -v "$global_svc_mon_mac > $lsp_mac" lsp_in.tcpdump], [1]) > > to fail because the tcpdump prints an extra line per packet with the > empty non existant inner Ethernet frame (it isn't really VXLAN traffic). > > change the invocation of tcpdump to include "-q" to disable payload > decoding for all protocols which since we are only looking at the MAC > addresses is fine. > > Signed-off-by: Jacob Tanenbaum <[email protected]> > > diff --git a/tests/system-ovn.at b/tests/system-ovn.at > index 3cb4b858a..e9e6d1ebd 100644 > --- a/tests/system-ovn.at > +++ b/tests/system-ovn.at > @@ -22353,8 +22353,8 @@ check ovn-nbctl --wait=hv sync > > dnl Check mac address of probe requests and replies, they should use the > dnl global svc_monitor_mac. > -NETNS_START_TCPDUMP([lsp], [-nne -i lsp -Q in 'udp port 42'], [lsp_in]) > -NETNS_START_TCPDUMP([lsp], [-nne -i lsp -Q out 'udp port 42 || icmp'], > [lsp_out]) > +NETNS_START_TCPDUMP([lsp], [-nneq -i lsp -Q in 'udp port 42'], [lsp_in]) > +NETNS_START_TCPDUMP([lsp], [-nneq -i lsp -Q out 'udp port 42 || icmp'], > [lsp_out]) > > dnl Start a backend server, the monitor should change status to "online". > NETNS_DAEMONIZE([lsp], [ncat -l -u 42], [nc.pid]) _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
