Bug 1568133
| Summary: | [Netvirt] DPDK VxLAN - Multicast traffic test has failed | ||
|---|---|---|---|
| Product: | Red Hat OpenStack | Reporter: | Ziv Greenberg <zgreenbe> |
| Component: | opendaylight | Assignee: | Victor Pickard <vpickard> |
| Status: | CLOSED DUPLICATE | QA Contact: | Itzik Brown <itbrown> |
| Severity: | high | Docs Contact: | |
| Priority: | high | ||
| Version: | 13.0 (Queens) | CC: | aadam, amuller, atelang, chrisw, fbaudin, mbabushk, mkolesni, nyechiel, sclewis, skramaja, srevivo, trozet, vpickard, yrachman, zgreenbe |
| Target Milestone: | beta | Keywords: | Triaged |
| Target Release: | 13.0 (Queens) | Flags: | zgreenbe:
needinfo-
zgreenbe: needinfo- |
| Hardware: | x86_64 | ||
| OS: | Linux | ||
| Whiteboard: | odl_netvirt | ||
| Fixed In Version: | Doc Type: | If docs needed, set a value | |
| Doc Text: | Story Points: | --- | |
| Clone Of: | Environment: |
N/A
|
|
| Last Closed: | 2018-04-26 07:39:15 UTC | Type: | Bug |
| Regression: | --- | Mount Type: | --- |
| Documentation: | --- | CRM: | |
| Verified Versions: | Category: | --- | |
| oVirt Team: | --- | RHEL 7.3 requirements from Atomic Host: | |
| Cloudforms Team: | --- | Target Upstream Version: | |
| Embargoed: | |||
|
Description
Ziv Greenberg
2018-04-16 19:33:49 UTC
Most of the drops here are in table 241 (Egress ACL) and table 81 (???). It would be helpful if you can provide the ofproto/trace on the node that is dropping the packets. You can do: 1) ovs-ofctl -O openflow13 show br-int (to get the ports) 2) ovs-appctl ofproto/trace br-int in_port=7,udp,dl_src=fa:16:3e:f4:bd:de,dl_dst=ff:ff:ff:ff:ff:ff,nw_dst=255.255.255.255,udp_dst=67,udp_src=68 ^replace with your packet/port info That way we can see which flows in the pipeline are being hit and which table is dropping the multicast. When using security groups with multicast traffic, you have to configure the port to allow ipv4 multicast address as shown below: openstack port set --allowed-address ip-address=226.94.1.1,mac-address=01:00:5e:5e:01:01 74ab3b8e-1b95-4fef-a60d-295856b714b6 Replace the port in the above command with the port under test, and please confirm if the multicast packets reach the receiver (listener1). (In reply to Victor Pickard from comment #3) > When using security groups with multicast traffic, you have to configure the > port to allow ipv4 multicast address as shown below: > > openstack port set --allowed-address > ip-address=226.94.1.1,mac-address=01:00:5e:5e:01:01 > 74ab3b8e-1b95-4fef-a60d-295856b714b6 > > > Replace the port in the above command with the port under test, and please > confirm if the multicast packets reach the receiver (listener1). Hi Victor, It did the trick, the multicast packets have been received in the listener1 instance as expected. My question is, how come when executing the same test but with neutron deployment instead, I didn't need to configure the openstack port at all. Thanks, Ziv Hi Ziv, With ODL as the backend driver, the multicast packets are dropped by acl checks. Configuring the port as above adds rules to the pipeline to allow packets that match the IP to egress the switch. It is my understanding that OVS (neutron deployment) will flood multicast packets to all ports (unless you have igmp snooping enabled, in which case only sent to registered listeners/receivers), and doesn't require explicit rules to allow packets to egress (based on your earlier comment). Do you think this is an issue, or just something that we need to make sure is documented properly when using ODL as backend driver? HI Victor, Thank you for a detailed explanation! I think the following question should be addressed to the product. Franck, Nir, please let us know what is your thoughts? *** This bug has been marked as a duplicate of bug 1550663 *** |