[OVSDB-268] LLDP Spoofing attack warning when using Openstack with ODL Cluster (both features) Created: 21/Jan/16 Updated: 11/Feb/16 Resolved: 11/Feb/16 |
|
| Status: | Resolved |
| Project: | ovsdb |
| Component/s: | openstack.net-virt |
| Affects Version/s: | unspecified |
| Fix Version/s: | None |
| Type: | Bug | ||
| Reporter: | Venkatrangan Govindarajan | Assignee: | Anil Vishnoi |
| Resolution: | Done | Votes: | 0 |
| Labels: | None | ||
| Remaining Estimate: | Not Specified | ||
| Time Spent: | Not Specified | ||
| Original Estimate: | Not Specified | ||
| Environment: |
Operating System: All |
||
| External issue ID: | 5040 |
| Priority: | High |
| Description |
|
When using Openstack and 3 node ODL cluster, with one ODL to handle neutron requests and the other two members as Managers in the OVS of Openstack. There were repeated warning messages in karf.log of one of the nodes ogress. |
| Comments |
| Comment by Venkatrangan Govindarajan [ 21/Jan/16 ] |
|
Used stable/beryllium code |
| Comment by Sam Hague [ 22/Jan/16 ] |
|
Anil, any idea what this lldp spoofing warning is? Is this a new feature? |
| Comment by Anil Vishnoi [ 22/Jan/16 ] |
|
Sam, Venkat, i never seen it in my environment, can you dump the rules on the OVS switches and see if there is drop rule installed or not. |
| Comment by Anil Vishnoi [ 02/Feb/16 ] |
|
venkat, i tried this locally with the same setup as yours and i am not seeing this issue. Can you please dump the output of "ovs-vsctl show" on all the nodes + flows installed on all the bridges on these nodes. |
| Comment by Venkatrangan Govindarajan [ 02/Feb/16 ] |
|
Anil, ODL1 -> Neutron Requests Still only ODL3 has the warning. |
| Comment by Venkatrangan Govindarajan [ 02/Feb/16 ] |
|
ovs-vsctl show control node: Bridge br-int Controller "tcp:11.12.13.5:6653" is_connected: true Controller "tcp:11.12.13.4:6653" is_connected: true fail_mode: secure Port "vxlan-12.14.16.3" Interface "vxlan-12.14.16.3" type: vxlan options: {key=flow, local_ip="12.14.16.2", remote_ip="12.14.16.3"} Port "tap1b35eb29-c3" Interface "tap1b35eb29-c3" Port br-int Interface br-int type: internal Port "tapc741c740-b3" Interface "tapc741c740-b3" Port "tap92b1f549-b3" Interface "tap92b1f549-b3" type: internal Port patch-ext Interface patch-ext type: patch options: {peer=patch-int} Port "tapf58200cd-00" Interface "tapf58200cd-00" type: internal ovs_version: "2.3.2" compute node: ubuntu@odl-os-compute-node:~/devstack$ sudo ovs-vsctl show bf460c25-5849-4874-9758-4fcbafbf9f5e Manager "tcp:11.12.13.5:6640" is_connected: true Manager "tcp:11.12.13.4:6640" is_connected: true Bridge br-ex Controller "tcp:11.12.13.5:6653" is_connected: true Controller "tcp:11.12.13.4:6653" is_connected: true fail_mode: secure Port "eth2" Interface "eth2" Port br-ex Interface br-ex type: internal Port patch-int Interface patch-int type: patch options: {peer=patch-ext} Bridge br-int Port "tap547166bc-d8" ovs_version: "2.3.2" |
| Comment by Venkatrangan Govindarajan [ 02/Feb/16 ] |
|
Control Node cookie=0x0, duration=3298.189s, table=0, n_packets=908, n_bytes=101696, dl_type=0x88cc actions=CONTROLLER:65535 br-ex |
| Comment by Venkatrangan Govindarajan [ 02/Feb/16 ] |
|
Compute Node br-ex has no flows installed. br-int has the lldp flows as needed |
| Comment by Venkatrangan Govindarajan [ 03/Feb/16 ] |
|
Other clues: Always happens in ODL3 when ODL2 and ODL3 are used as OVS Managers and Controllers for testing. the warning occurs every second. ODL3 is the owner for all the entities, I think since br-ex is connected to public network, Maybe that is causing this error. , { "name": "member-2" } ], , { "name": "member-2" } ], , , { "name": "member-1" } ], , { "name": "member-2" } ], , { "name": "member-2" } ], , { "name": "member-2" } ], , , { "name": "member-2" } ], ] , , { "name": "member-1" } ], |
| Comment by Venkatrangan Govindarajan [ 03/Feb/16 ] |
|
(In reply to Venkatrangan Govindarajan from comment #8) Will need to add this as a seperate bug |
| Comment by Anil Vishnoi [ 10/Feb/16 ] |
|
In clustered environment, if two devices are owner by two different controller, LLDP packet sent by one controller node will go to another controller node through it's connected device. Currently we have a fix for security vulnerability fix that check if that LLDP packet was send by the same controller node and if not, it will spit the warning message reported in this bug. In 3 node cluster that's a very valid scenario, because different devices are connected to different controller node, but because of the existing security fix, it will spit warning message because that packet was not sent by the same controller. This requires a fix at the openflowplugin application level and it's not an easy fix from security perspective. Net-virt don't usse any of the topology information because it's end host based solution, so we don't really want lldp speaker to send lldp packets to the switch. I disabled the lldp speaker for net-virt using the following patch |