yahoo-eng-team team mailing list archive

[Alexey Kashavkin <2028131@xxxxxxxxxxxxxxxxxx>]

Public bug reported:

It happens only with virtual machines in the same provider network
without any security groups on their ports on the same compute node.
Packets are not sending to virtual machines in this network on another
compute nodes, but for such network in ovn-trace you can see that this
packets must go to another compute into ports of the same network to,
but it wasn't happening. (I didn't save output of trace, take my word
for it.)

It independent by compute node or provider network. It might different
provider network without port security enabled and different compute
node. The main it happens as I wrote above when VMs in the same compute
node and in the same network.

You can see all information in attachment file.

I use TripleO RDO Train version with tag for rpm packages and containers - current-tripleo-rdo.
OVN 2.12.90 (DB Schema 5.16.0)
neutron-15.3.5-0.20220117160409
openvswitch2.15-2.15.0-94 on host
Host OS is Rocky Linux 8.5

But I've seen it happen the same way on RHOSP 16.1 with OVN-2.13-20.06.1
and Neutron v15.1.1-0.20200611111910

** Affects: neutron
     Importance: Undecided
         Status: New


** Tags: ovn

** Attachment added: "All information can be seen here"
   https://bugs.launchpad.net/bugs/2028131/+attachment/5687104/+files/multicast_issue.txt

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to neutron.
https://bugs.launchpad.net/bugs/2028131

Title:
  [ovn] All traffic duplicates to tap-interfaces on the same provider
  network without port security enabled on the same compute node.

Status in neutron:
  New

Bug description:
  It happens only with virtual machines in the same provider network
  without any security groups on their ports on the same compute node.
  Packets are not sending to virtual machines in this network on another
  compute nodes, but for such network in ovn-trace you can see that this
  packets must go to another compute into ports of the same network to,
  but it wasn't happening. (I didn't save output of trace, take my word
  for it.)

  It independent by compute node or provider network. It might different
  provider network without port security enabled and different compute
  node. The main it happens as I wrote above when VMs in the same
  compute node and in the same network.

  You can see all information in attachment file.

  I use TripleO RDO Train version with tag for rpm packages and containers - current-tripleo-rdo.
  OVN 2.12.90 (DB Schema 5.16.0)
  neutron-15.3.5-0.20220117160409
  openvswitch2.15-2.15.0-94 on host
  Host OS is Rocky Linux 8.5

  But I've seen it happen the same way on RHOSP 16.1 with
  OVN-2.13-20.06.1 and Neutron v15.1.1-0.20200611111910

To manage notifications about this bug go to:
https://bugs.launchpad.net/neutron/+bug/2028131/+subscriptions