Public bug reported:

I'm facing a problem in a customer where a VM plugged to a bridge will not get 
DHCP responses from dhcp server running on the host or externally (the bridge 
has level 2 external uplink).

Relevant information:

- focal with latest GA kernel (5.4.0-88)
- tested with HWE (5.11) and same results
- bridge is configured using netplan, with bond as uplink, ip on the bridge
- removed bond from the equation (bridge --> eno1), same result
- host can access external network with no problems using the uplink plugged in 
the bridge
- dhcp server (ISC-DHCP) configured on the host
- stp on the bridge is disabled

Create VM, attach it to the bridge, power it on, boots. Requests IP via
DHCP. The discover can be seen on tcpdump inside the vm, and also on the
tcpdump on the host bridge. DHCP server answers with Offer, that packet
can be seen in the bridge but do not enter the VM.

If I manually put an IP on the VM, it pings the host and the external
world. Mac/ARP resolution seems to be working both ways (arping tested).

Now, if I UNPLUG the uplink from the bridge (leaving only the host
itself with IP on the bridge and one VM plugged on it) then the VM can
get IP via DHCP, the Offer comes in. I'm not sure why it matters having
the uplink when dhcp requests are being answered locally.

After being able to ping externally, if I remove the IP and try DHCP
again, it fails again.

The hardware is a IBM HS23 blade with Emulex 10Gbit cards:

ubuntu@app2maas001p:~$ lspci -v | grep -i net
16:00.0 Ethernet controller: Emulex Corporation OneConnect 10Gb NIC (be3) (rev 
03)
        Kernel driver in use: be2net
        Kernel modules: be2net
16:00.1 Ethernet controller: Emulex Corporation OneConnect 10Gb NIC (be3) (rev 
03)
        Kernel driver in use: be2net
        Kernel modules: be2net

I'm verifying if there is firmware updates available but I doubt it. No
extra features (like FCoE, partitioning, etc.) are enabled on the card.
It's a single dual-port mezanine card.

** Affects: linux (Ubuntu)
     Importance: Undecided
         Status: New

-- 
You received this bug notification because you are a member of Kernel
Packages, which is subscribed to linux in Ubuntu.
https://bugs.launchpad.net/bugs/1946028

Title:
  DHCP Offer not traversing linux bridge

Status in linux package in Ubuntu:
  New

Bug description:
  
  I'm facing a problem in a customer where a VM plugged to a bridge will not 
get DHCP responses from dhcp server running on the host or externally (the 
bridge has level 2 external uplink).

  Relevant information:

  - focal with latest GA kernel (5.4.0-88)
  - tested with HWE (5.11) and same results
  - bridge is configured using netplan, with bond as uplink, ip on the bridge
  - removed bond from the equation (bridge --> eno1), same result
  - host can access external network with no problems using the uplink plugged 
in the bridge
  - dhcp server (ISC-DHCP) configured on the host
  - stp on the bridge is disabled

  Create VM, attach it to the bridge, power it on, boots. Requests IP
  via DHCP. The discover can be seen on tcpdump inside the vm, and also
  on the tcpdump on the host bridge. DHCP server answers with Offer,
  that packet can be seen in the bridge but do not enter the VM.

  If I manually put an IP on the VM, it pings the host and the external
  world. Mac/ARP resolution seems to be working both ways (arping
  tested).

  Now, if I UNPLUG the uplink from the bridge (leaving only the host
  itself with IP on the bridge and one VM plugged on it) then the VM can
  get IP via DHCP, the Offer comes in. I'm not sure why it matters
  having the uplink when dhcp requests are being answered locally.

  After being able to ping externally, if I remove the IP and try DHCP
  again, it fails again.

  The hardware is a IBM HS23 blade with Emulex 10Gbit cards:

  ubuntu@app2maas001p:~$ lspci -v | grep -i net
  16:00.0 Ethernet controller: Emulex Corporation OneConnect 10Gb NIC (be3) 
(rev 03)
        Kernel driver in use: be2net
        Kernel modules: be2net
  16:00.1 Ethernet controller: Emulex Corporation OneConnect 10Gb NIC (be3) 
(rev 03)
        Kernel driver in use: be2net
        Kernel modules: be2net

  I'm verifying if there is firmware updates available but I doubt it.
  No extra features (like FCoE, partitioning, etc.) are enabled on the
  card. It's a single dual-port mezanine card.

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1946028/+subscriptions


-- 
Mailing list: https://launchpad.net/~kernel-packages
Post to     : kernel-packages@lists.launchpad.net
Unsubscribe : https://launchpad.net/~kernel-packages
More help   : https://help.launchpad.net/ListHelp

Reply via email to