Re: [ovs-dev] [PATCH net] openvswitch: Set the skbuff pkt_type for proper pmtud support.

[Aaron Conole]

Eelco Chaudron <echaudro@xxxxxxxxxx> writes:

> On 25 Mar 2024, at 13:37, Ilya Maximets wrote:
>
>> On 3/25/24 13:22, Aaron Conole wrote:
>>> Eelco Chaudron <echaudro@xxxxxxxxxx> writes:
>>>
>>>> On 22 Mar 2024, at 20:06, Aaron Conole wrote:
>>>>
>>>>> Open vSwitch is originally intended to switch at layer 2, only dealing with
>>>>> Ethernet frames.  With the introduction of l3 tunnels support, it crossed
>>>>> into the realm of needing to care a bit about some routing details when
>>>>> making forwarding decisions.  If an oversized packet would need to be
>>>>> fragmented during this forwarding decision, there is a chance for pmtu
>>>>> to get involved and generate a routing exception.  This is gated by the
>>>>> skbuff->pkt_type field.
>>>>>
>>>>> When a flow is already loaded into the openvswitch module this field is
>>>>> set up and transitioned properly as a packet moves from one port to
>>>>> another.  In the case that a packet execute is invoked after a flow is
>>>>> newly installed this field is not properly initialized.  This causes the
>>>>> pmtud mechanism to omit sending the required exception messages across
>>>>> the tunnel boundary and a second attempt needs to be made to make sure
>>>>> that the routing exception is properly setup.  To fix this, we set the
>>>>> outgoing packet's pkt_type to PACKET_OUTGOING, since it can only get
>>>>> to the openvswitch module via a port device or packet command.
>>>>
>>>> Is this not a problem when the packet comes from the bridge port in the kernel?
>>>
>>> It very well may be an issue there as well, but the recommendation is to
>>> operate with the bridge port down as far as I know, so I don't know if
>>> this issue has been observed happening from the bridge port.
>>
>> FWIW, bridge ports are typically used as an entry point for tunneled
>> traffic so it can egress from a physical port attached to OVS.  It means
>> they are pretty much always UP in most common setups like OpenStack or
>> ovn-kubernetes and handle a decent amount of traffic.  They are also used
>> to direct some other types of traffic to the host kernel.
>
> +1 here, I’m talking about the same port. I think we only advise
> having this down for userspace bridges, but not in the case the bridge
> is the tunnel endpoint.

Okay, I'll confirm about up/down, but it seems like it shouldn't matter
and we should be setting the outgoing type.

>> Unless I misunderstood which ports we're talking about here.
>>
>> Best regards, Ilya Maximets.