Manuel, Might be a dumb question, but are you making sure no OVS instances connect to ODL until netvirt is fully up (aka can curl restconf/operational/network-topology:network-topology/topology/netvirt:1) and the tables have been offset? Seems like this could be a timing problem there.
Tim Rozet Red Hat SDN Team ----- Original Message ----- From: "Srikanth Lingala" <[email protected]> To: "Manuel Buil" <[email protected]>, [email protected], "Brady Allen Johnson" <[email protected]>, "Sam Hague" <[email protected]> Cc: [email protected], [email protected] Sent: Monday, March 6, 2017 1:12:38 PM Subject: Re: [opnfv-tech-discuss] [sfc-dev] [odl-sfc] Bug in OPNFV-SFC test cases Hi, Sometimes, I also encountered the same problem. I think some issue is there in the ODL. I am using OPNFV Colorado 3.0. VM's will not get IP address from Neutron DHCP agent. To fix the issue, I needed to do the following: ovs-vsctl del-br br-int ovs-vsctl del-manager service neutron-server restart service opendaylight restart ovs-vsctl set-manager tcp:<ODL_IP>:6640 Sometimes, SFC tables are adding from table id's 11, but not table id 151. After restarting ODL, that issue fixed for me. Regards, Srikanth. -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Manuel Buil Sent: Monday, March 06, 2017 11:12 PM To: [email protected]; Brady Allen Johnson <[email protected]>; Sam Hague <[email protected]> Cc: [email protected]; [email protected] Subject: [sfc-dev] [odl-sfc] Bug in OPNFV-SFC test cases Hello guys, We moved to ODL Boron SR2 lately and the most basic OPNFV test cases are failing because VMs don't get an IP lease from the DHCP server. When checking the tables of the compute nodes, they seem a bit messy to me and it looks like that perhaps a patch might have broken the compatibility with the way we are currently separating tables between SFC and Netvirt until SFC is integrated with Genius (remember we are configuring the variables: table-offset:1 and sfc-of-table-offset:150, sfc-of-app-egress-table-offset:11). Here are the flows: http://pastebin.com/Bii0vaNC As you can see, all packets are dropped in table 110 (those are DHCP packets). There are strange things like for example: there is no way to reach table=1, table=21, table=41,etc. Exactly the tables which are needed and modified by the configuration of the offset +1, that's why I suspect that there was a patch in netvirt which might have not taken this offset into account. Do you agree that the tables are wrong? Any idea what can be causing that flow configuration? Thanks, Manuel _______________________________________________ sfc-dev mailing list [email protected] https://lists.opendaylight.org/mailman/listinfo/sfc-dev _______________________________________________ opnfv-tech-discuss mailing list [email protected] https://lists.opnfv.org/mailman/listinfo/opnfv-tech-discuss _______________________________________________ sfc-dev mailing list [email protected] https://lists.opendaylight.org/mailman/listinfo/sfc-dev
