Skip to content

Commit

Permalink
Apply suggestions from code review
Browse files Browse the repository at this point in the history
Co-authored-by: Billy McFall <22157057+Billy99@users.noreply.github.com>
  • Loading branch information
pliurh and Billy99 authored Sep 16, 2021
1 parent 16af993 commit d052542
Showing 1 changed file with 6 additions and 6 deletions.
12 changes: 6 additions & 6 deletions enhancements/network/dpu-ovnkube-operator.md
Original file line number Diff line number Diff line change
Expand Up @@ -25,15 +25,15 @@ status: implementable

## Summary

To facilitate the management of Nvidia BlueField-2 DPU, a two-cluster design is
proposed. Under such design, a BlueFiled-2 card will be provisioned as a worker
To facilitate the management of Nvidia BlueField-2 DPU, a two-cluster design is being
proposed. Under such design, a BlueField-2 card will be provisioned as a worker
node of the ARM-based infra cluster, whereas the tenant cluster where the normal
user applications run on, is composed of the X86 servers.

The OVN-Kubernetes components are spread over the two clusters. On the tenant
cluster side, the Cluster Network Operator is in charge of the management of the
ovn-kube components. On the infra cluster side, we propose to create a new
operator to be responsible for the life-cycle management of ovn-kube components
operator to be responsible for the life-cycle management of the ovn-kube components
and the necessary host network initialization on DPU cards.

## Motivation
Expand All @@ -50,7 +50,7 @@ with each other.
### Goals

- Design a new operator that
- manages the ovn-kube components that serve the tenant cluster.
- manages the ovn-kube components running on the DPU that serve the tenant cluster.
- enable the switchdev mode on DPUs.
- add PF representor to the br-ex bridge, so that the x86 host can connect to
the cluster network of the tenant cluster.
Expand Down Expand Up @@ -101,7 +101,7 @@ configuration on the DPUs which we want to enable hardware offloading:

- Add PF representor to the br-ex bridge
- Enable switchdev mode on DPU nodes
- Enable OVS hareware offloading: other_config:hw-offload=true
- Enable OVS hardware offloading: other_config:hw-offload=true
- Apply udev rules for BF2 interfaces

We assume a custom MachineConfigPool shall be created for all the DPUs, and the
Expand All @@ -112,7 +112,7 @@ service to do the aforementioned configuration.
Example configurations:

```yaml
apiVersion: smartnic.openshift.io/v1
apiVersion: dpu.openshift.io/v1
kind: OVNKubeConfig
metadata:
name: cluster
Expand Down

0 comments on commit d052542

Please sign in to comment.