Yoav The OVS version you are using should be OK and there is no need for extra patch anywhere that I am aware of right.
Please, check that the both switches are properly connected to opendaylight openflow controller (`ovs-vsctl show` will show it) and that the service-node used in the configuration matches the openflow node id in the ODL openflow inventory. BR Jaime. -----Original Message----- From: Yoav Cohen <[email protected]> To: opendaylight sfc <[email protected]>, [email protected] Subject: RE: [sfc-dev] SFC Tutorial Issues Date: Mon, 11 Mar 2019 16:34:56 +0000 Hi Jaime, Thanks for the update. I now understand the limitations WRT OVS 2.6.1 and ODL Oxygen. I haven't tested this combination, since I thought that any version of OVS *starting from* version 2.6.1 would be ok. But now I understand that for ODL with Oxygen version it needs a special patch of OVS (i.e. OVS version 2.6.1). WRT ODL with Fluorine and OVS version 2.10.1: copying from my email below where I wrote "When running the same scripts, using ODL Fluorine version and viewing the results in the mininet based OF switches, switch #2 is not configured at all, while switch #3 seems to be configured correctly." So it seems that one switch (switch#3) is configured correctly, while the other switch (switch#2) is not configured at all. Any comments on this situation?. Do I need to use OVS version 2.9 (instead of 2.10.1, which I used) or maybe a higher version of OVS? Or is there a patch that needs to be inserted into the scripts to correct this situation? Thanks, Yoav. -----Original Message----- From: Jaime Caamaño Ruiz <[email protected]> Sent: Monday, March 11, 2019 5:30 PM To: Yoav Cohen <[email protected]>; opendaylight sfc < [email protected]> Subject: Re: [sfc-dev] SFC Tutorial Issues Hello Yoav You can check the SFC release notes of both Oxygen [1] and Fluorine [2]. On Oxygen, it states > SFC needs changes in OVS to include the Network Service Headers > (NSH) > Chaining encapsulation feature. This patch has been ongoing for > quite > a while, but has finally been officially merged in OVS 2.8. > OpenDaylight will be updated to use this new version of OVS in the > Fluorine release. Until then, SFC will use a branched version of OVS > based on 2.6.1, called the “Yi Yang Patch”, located here. Previous > versions of this OVS patch only supported VXLAN-GPE + NSH > encapsulation, but this version supports both ETH + NSH and VXLAN-GPE > + ETH + NSH. On Fluorine > Support for Network Service Headers (NSH) has been officially > implemented in the OVS 2.9 release. As a result, support for the > older, unofficial branched OVS versions has been removed, and now > only > OVS 2.9 is supported. So OVS version 2.10.1 will only work with SFC at ODL release Fluorine or newer. [1] https://docs.opendaylight.org/en/stable-oxygen/release-notes/projec ts/sfc.html [2] https://docs.opendaylight.org/en/stable-fluorine/release-notes/proj ects/sfc.html BR Jaime. -----Original Message----- From: Yoav Cohen <[email protected]> To: opendaylight sfc <[email protected]>, [email protected] Subject: [sfc-dev] SFC Tutorial Issues Date: Fri, 22 Feb 2019 12:57:09 +0000 Hi All, I’m running the tutorial scripts for NSH listed here https://docs.opend aylight.org/en/stable-oxygen/user-guide/service-function-chaining.html I have upgraded OVS to version 2.10.1 When running the scripts using ODL Oxygen version and view the results in the mininet based OF switches 2&3 I get the following results: $ sudo ovs-ofctl -O OpenFlow13 dump-flows s3 cookie=0x14, duration=755.325s, table=0, n_packets=306, n_bytes=26098, priority=5 actions=goto_table:1 cookie=0x14, duration=755.325s, table=1, n_packets=306, n_bytes=26098, priority=5 actions=drop cookie=0x14, duration=755.325s, table=2, n_packets=0, n_bytes=0, priority=5 actions=goto_table:3 cookie=0x14, duration=755.325s, table=3, n_packets=0, n_bytes=0, priority=5 actions=goto_table:4 cookie=0x14, duration=755.325s, table=4, n_packets=0, n_bytes=0, priority=5 actions=goto_table:10 cookie=0x14, duration=755.325s, table=10, n_packets=0, n_bytes=0, priority=5 actions=drop This does not seem to be correct. Also it does not correspond to the results specified in the documentation. For example, for table=1 the documentation notes: Priority Match Action 256 EtherType==0x8847 (MPLS unicast) Goto Table 2 256 EtherType==0x8100 (VLAN) Goto Table 2 256 EtherType==0x0800,udp,tp_dst==4789 (IP v4) Goto Table 2 5 Match Any Drop I would have expected the OF switch configuration above to match the 3rd row, but instead it seems to match the 4th row. When running the same scripts, using ODL Fluorine version and view the results in the mininet based OF switches, switch #2 is not configured at all, while switch #3 seems to be configured correctly. Can someone clarify the different behaviors using Oxygen & Fluorine versions and what needs to be done to receive the correct results in either of the ODL versions. Thanks in advance, Yoav Cohen. _______________________________________________ sfc-dev mailing list [email protected] https://lists.opendaylight.org/mailman/listinfo/sfc-dev _______________________________________________ sfc-dev mailing list [email protected] https://lists.opendaylight.org/mailman/listinfo/sfc-dev
