# Offloading flowtable, LAG and VLAN

**URL:** <https://forums.developer.nvidia.com/t/offloading-flowtable-lag-and-vlan/290203>\
**Category:** Ethernet Adapter Cards\
**Created:** [April 19, 2024, 7:03am UTC](https://forums.developer.nvidia.com/t/offloading-flowtable-lag-and-vlan/290203 "2024-04-19T07:03:51Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![flo-hm](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@flo-hm](https://forums.developer.nvidia.com/u/flo-hm)\
**Post date:** [April 19, 2024, 7:03am UTC](https://forums.developer.nvidia.com/t/offloading-flowtable-lag-and-vlan/290203/1 "2024-04-19T07:03:51Z")

</div>

Hello together,

we are currently trying to get a mixture of Hardware LAG, VLAN Tagging and Flowtable Offloading working on our debian12 system using a ConnectX-6 Lx

Creating the HW Lag using the card running in switchdev works like charm.

Creating the VLANs and performing some filtering afterwards gets a bit more complicated. I tried to simply create a tagged version of the vlan and tell the nftable rules to perform some filtering. Thats working fine but seems to run in the kernel/software instead of getting ofloaded to the hardware since the system is running on full load handling the interrupts.

Is there any (recommended) way of building such a setup. Should I use several virtual functions and tag them?

Thanks in advance!

---

<div class="post-metadata">

**Author:** ![sribhargavid](https://sea2.discourse-cdn.com/nvidia/user_avatar/forums.developer.nvidia.com/sribhargavid/32/367020_2.png) [@sribhargavid](https://forums.developer.nvidia.com/u/sribhargavid)\
**Post date:** [April 19, 2024, 9:29pm UTC](https://forums.developer.nvidia.com/t/offloading-flowtable-lag-and-vlan/290203/2 "2024-04-19T21:29:28Z")

</div>

Hello @flo-hm,

Thank you for posting your query on our community. Please refer to this article on configuring VLAN offload using tc and let us know if it helps - [ESPCommunity](https://enterprise-support.nvidia.com/s/?type=article&id=Configuring-VLAN-Push-Pop-Offload-Using-tc)

Thanks,  
Bhargavi

---

<div class="post-metadata">

**Author:** ![flo-hm](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@flo-hm](https://forums.developer.nvidia.com/u/flo-hm)\
**Post date:** [April 20, 2024, 6:42pm UTC](https://forums.developer.nvidia.com/t/offloading-flowtable-lag-and-vlan/290203/3 "2024-04-20T18:42:19Z")

</div>

Hi,

the link only points to the starting page of the Enterprise Support Portal.

In the best case I want to offload firewall rules using nftables in combination with flowtables and the offload flag

My vision would be to define a flowtable using something like that:  
flowtable f {

hook ingress priority 0; devices = { bond0.123, bond0.124 };

flags offload;

}

where bond0.123 and bond0.124 are a tagged version of a LACP Bond using the HW LAG functionality of my mellanox cards running in switchdev mode.

To sum up I would like to have several vlans → one bond → mellanox card

From my first tests offloading to the mellanox card it sellf seems to work but the bond in between seems to cause some problems…