yahoo-eng-team team mailing list archive

[Milan <groshevmilan@xxxxxxxxx>]

Public bug reported:

On the virtual machines we have two network interfaces. Eth0
(10.70.70.0/24) and eth1 (10.71.71.0/24)

i use eth1 for managment and eth0 for public access

The DHCP reply packets are not able to reach the TAP interface on the
compute node because the bridge is sending the packet back to eth0.

Eth0 is saying to the bridge that he has the INSTANCE interface and the
packet is sent back to eth0. Eth0 sends it again to br interface and
they get in a loop. At the end the packet gets droped. Here brctl
showmacs on the bridge

i am using only linux bridge no OVS

here is brctl showmacs form the bridge of the compute node

$ brctl showmacs brq466a96cb-7a
port no mac addr                is local?       ageing timer
  2     00:0c:29:18:5d:36       no                 0.26
  2     00:0c:29:a0:61:0a       yes                0.00
  2     00:0c:29:a0:61:0a       yes                0.00
  2     00:19:56:be:2d:ad       no                 1.85
  2     00:19:56:be:2d:ae       no                 1.85
  2     64:9e:f3:35:b0:37       no                 0.00
  2     fa:16:3e:76:b2:65       no                18.17
  2     fa:16:3e:f4:33:9c       no                11.16
  1     fe:16:3e:76:b2:65       yes                0.00
  1     fe:16:3e:76:b2:65       yes                0.00

The interfaces are in promiscus mode! i can not understand why eth0 is
registering the instance on his port, port 2 !

i use tcpdump to verify the presence of the DHCP reply and it is
arriving to the bridge

** Affects: neutron
     Importance: Undecided
         Status: New

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to neutron.
https://bugs.launchpad.net/bugs/1552708

Title:
  OpenStack Compute node linux bridge error in configuration

Status in neutron:
  New

Bug description:
  On the virtual machines we have two network interfaces. Eth0
  (10.70.70.0/24) and eth1 (10.71.71.0/24)

  i use eth1 for managment and eth0 for public access

  The DHCP reply packets are not able to reach the TAP interface on the
  compute node because the bridge is sending the packet back to eth0.

  Eth0 is saying to the bridge that he has the INSTANCE interface and
  the packet is sent back to eth0. Eth0 sends it again to br interface
  and they get in a loop. At the end the packet gets droped. Here brctl
  showmacs on the bridge

  i am using only linux bridge no OVS

  here is brctl showmacs form the bridge of the compute node

  $ brctl showmacs brq466a96cb-7a
  port no mac addr                is local?       ageing timer
    2     00:0c:29:18:5d:36       no                 0.26
    2     00:0c:29:a0:61:0a       yes                0.00
    2     00:0c:29:a0:61:0a       yes                0.00
    2     00:19:56:be:2d:ad       no                 1.85
    2     00:19:56:be:2d:ae       no                 1.85
    2     64:9e:f3:35:b0:37       no                 0.00
    2     fa:16:3e:76:b2:65       no                18.17
    2     fa:16:3e:f4:33:9c       no                11.16
    1     fe:16:3e:76:b2:65       yes                0.00
    1     fe:16:3e:76:b2:65       yes                0.00

  The interfaces are in promiscus mode! i can not understand why eth0 is
  registering the instance on his port, port 2 !

  i use tcpdump to verify the presence of the DHCP reply and it is
  arriving to the bridge

To manage notifications about this bug go to:
https://bugs.launchpad.net/neutron/+bug/1552708/+subscriptions