yahoo-eng-team team mailing list archive

[Alec Hothan <ahothan@xxxxxxxxx>]

Public bug reported:

Testbed configuration to reproduce:
- 1 TOR with 1 or 2 compute nodes
- 1 TOR with 1 network node
- 1 Spine connecting the 2 TOR

data path: VM1->TOR1->Spine->TOR2->router->TOR2->Spine->TOR1->VM2
VM1 and VM2 are on different networks
The TOR are Nexus 9K (C9396PX NXOS 6.1)

Ping from VM1 to VM2 using fixed IP fails 100% of the time.
The test was performed with VM1 and VM2 running on different compute nodes under TOR1. It will also probably fail if they run on the same compute node under TOR1.

Other data paths where ping is working:
VM1->TOR->router->TOR->VM2 (VM1 and VM2 are in the same rack as the network node and run on the same compute node)
VM1->TOR1->Spine->TOR2->router->TOR2->VM2 (VM2 is on the same rack as the network node)

When using floating IPs (ie going through NAT) ping on:
- VM in same rack as network node is OK: VM1->TOR->router->->TOR->default gateway->TOR->VM2
- at least one VM in a different rack than network node fails

These issues can be reproduced manually using ping or in an automated
way using the VMTP tool.

Versioning:
https://github.com/cisco-openstack/neutron.git (staging/kiloplus branch)
openstack and devstack: stable/juno

** Affects: neutron
     Importance: Undecided
         Status: New


** Tags: cisco ml2

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to neutron.
https://bugs.launchpad.net/bugs/1412903

Title:
  Cisco Nexus VxLAN ML2: ping fails with fixed IP with remote Network
  Node

Status in OpenStack Neutron (virtual network service):
  New

Bug description:
  Testbed configuration to reproduce:
  - 1 TOR with 1 or 2 compute nodes
  - 1 TOR with 1 network node
  - 1 Spine connecting the 2 TOR

  data path: VM1->TOR1->Spine->TOR2->router->TOR2->Spine->TOR1->VM2
  VM1 and VM2 are on different networks
  The TOR are Nexus 9K (C9396PX NXOS 6.1)

  Ping from VM1 to VM2 using fixed IP fails 100% of the time.
  The test was performed with VM1 and VM2 running on different compute nodes under TOR1. It will also probably fail if they run on the same compute node under TOR1.

  Other data paths where ping is working:
  VM1->TOR->router->TOR->VM2 (VM1 and VM2 are in the same rack as the network node and run on the same compute node)
  VM1->TOR1->Spine->TOR2->router->TOR2->VM2 (VM2 is on the same rack as the network node)

  When using floating IPs (ie going through NAT) ping on:
  - VM in same rack as network node is OK: VM1->TOR->router->->TOR->default gateway->TOR->VM2
  - at least one VM in a different rack than network node fails

  These issues can be reproduced manually using ping or in an automated
  way using the VMTP tool.

  Versioning:
  https://github.com/cisco-openstack/neutron.git (staging/kiloplus branch)
  openstack and devstack: stable/juno

To manage notifications about this bug go to:
https://bugs.launchpad.net/neutron/+bug/1412903/+subscriptions