Enable SRIOV - Kernel Driver

SRIOV and DPDK

SRI-IOV WorkerNode Prerequisites

In order for the cni-sriov-plugin to start reading Virtual Functions resources as well HostNetworkTemplate objects can configure virtual functions you need to perform the following steps.

  1. Enable** VT-D** at BIOS level.

  2. Enable SRIOV feature globally and/or per nic according to the nic you are working with, for example Intel x710 should be enabled per nic at BIOS level.

  3. Upgrade Kernel boot line to enable intel_iommu=on and iommu=pt

$ grubby --update-kernel=ALL --args="iommu=pt intel_iommu=on"
$ reboot

Huge page Support

Most network applications use huge pages, so you may want to enable that. Please edit /etc/default/grub and add huge-pages

GRUB_CMDLINE_LINUX="nofb nomodeset vga=normal iommu=pt intel_iommu=on default_hugepagesz=1G hugepagesz=1G hugepages=16"
#Rebuild grub.cfg
grub2-mkconfig -o /boot/grub2/grub.cfg && reboot

Create SR-IOV ConfigMap

SR-IOV ConfigMap

Please copy paste the following definition update the values of your Physical Functions and ranges or Virtual Functions as well as Drivers and apply it to our cluster, in case that you have multiple nics, that each one uses different kernels drivers please omit the drivers section and just mentioned pfName within the selectors section.

SR-IOV - Kernel Driver ConfigMap

apiVersion: v1
kind: ConfigMap
metadata:
name: sriovdp-config
namespace: default
data:
config.json: |
{
"resourceList": [{
"resourceName": "intel_sriov_kernel0",
"selectors": {
"pfNames": ["em1#0-6"],
"drivers": ["igbvf"]
}
}
]
}

Create Network Attach Definition SR-IOV type

Please copy paste the following definition update the values of your subnet and apply it to our cluster:

apiVersion: "k8s.cni.cncf.io/v1"
kind: NetworkAttachmentDefinition
metadata:
name: sriov-kernelnet0
annotations:
k8s.v1.cni.cncf.io/resourceName: intel.com/intel_sriov_kernel0
spec:
config: '{
"type": "sriov",
"cniVersion": "0.3.1",
"name": "sriov-kernelnet0",
"spoofchk": "off",
"type": "sriov",
"vlan": 80,
"ipam": {
"type": "whereabouts",
"range": "192.168.80.0/24",
"range_start": "192.168.80.20",
"range_end": "192.168.80.50",
"gateway": "192.168.80.1"
}
}'

The master key value is the reference to the *_second nic *_in our worker nodes.

# kubectl apply -f networkattachdefinition-sriov.yaml

Network Attach Definition Validation SR-IOV type

Let’s validate our work by listing and describing our new Network Attach Definition

$ kubectl get net-attach-def
NAME AGE
sriov-net-a 5d21h
$ kubectl describe net-attach-def sriov-kernelnet0
Name: sriov-kernelnet0
Namespace: default
Labels: <none>
Annotations: k8s.v1.cni.cncf.io/resourceName: intel.com/intel_sriov_kernel0
kubectl.kubernetes.io/last-applied-configuration:
{"apiVersion":"k8s.cni.cncf.io/v1","kind":"NetworkAttachmentDefinition","metadata":{"annotations":{"k8s.v1.cni.cncf.io/resourceName":"inte...
API Version: k8s.cni.cncf.io/v1
Kind: NetworkAttachmentDefinition
Metadata:
Creation Timestamp: 2020-09-23T18:53:31Z
Generation: 1
Resource Version: 174547
Self Link: /apis/k8s.cni.cncf.io/v1/namespaces/default/network-attachment-definitions/sriov-kernelnet0
UID: c6fd410a-97d8-42d4-86d5-56ef447e1dfa
Spec:
Config: { "type": "sriov", "name": "sriov-kernelnet0", "spoofchk": "off", "type": "sriov", "vlan": 80, "ipam": { "type": "whereabouts", "range": "192.168.80.0/24", "range_start": "192.168.80.20", "range_end": "192.168.80.50", "gateway": "192.168.80.1" } }

Create Pods with SR-IOV interfaces (Kernel-Driver)

$ cat pod0-case3.yaml
apiVersion: v1
kind: Pod
metadata:
name: pod0-case-03-sriov-kernel
annotations:
k8s.v1.cni.cncf.io/networks: sriov-kernelnet0
spec:
containers:
- name: pod0-case-03
image: docker.io/centos/tools:latest
command:
- /sbin/init
resources:
requests:
intel.com/intel_sriov_kernel0: '1'
limits:
intel.com/intel_sriov_kernel0: '1'

Deploy the new pods

$ kubectl apply -f pod0-case3.yaml

Validate Pods Creation with SR-IOV interfaces (Kernel Driver)

Let’s validate your work by confirming that the pods got created with an additional interface by doing the following commands:

$ kubectl get pods -o wide
NAME READY STATUS RESTARTS AGE IP NODE NOMINATED NODE READINESS GATES
pod0-case-02 1/1 Running 0 66m 10.135.1.147 192.168.50.14 <none> <none>
pod1-case-02 1/1 Running 0 62m 10.135.1.148 192.168.50.14 <none> <none>
$ kubectl exec -it pod0-case-02 -- ip -d address
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1
link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 promiscuity 0 numtxqueues 1 numrxqueues 1
inet 127.0.0.1/8 scope host lo
valid_lft forever preferred_lft forever
3: eth0@if25: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1472 qdisc noqueue state UP group default
link/ether de:ff:c2:57:c6:e7 brd ff:ff:ff:ff:ff:ff link-netnsid 0 promiscuity 0
veth numtxqueues 1 numrxqueues 1
inet 10.135.1.147/24 brd 10.135.1.255 scope global eth0
valid_lft forever preferred_lft forever
4: net1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UNKNOWN group default
link/ether fa:16:3e:10:31:25 brd ff:ff:ff:ff:ff:ff promiscuity 0
ipvlan mode l2 numtxqueues 1 numrxqueues 1
inet 192.168.80.20/24 brd 192.168.70.255 scope global net1

SR-IOV VF Testing

SR-IOV Kernel Driver Validation

$ kubectl exec pod0-case-03 -- ethtool -i net2
driver: igbvf
version: 2.4.0-k
firmware-version:
expansion-rom-version:
bus-info: 0000:01:10.2
supports-statistics: yes
supports-test: yes
supports-eeprom-access: no
supports-register-dump: yes
supports-priv-flags: no
$ kubectl exec pod0-case-03 -- ethtool -i net1
driver: igbvf
version: 2.4.0-k
firmware-version:
expansion-rom-version:
bus-info: 0000:01:10.0
supports-statistics: yes
supports-test: yes
supports-eeprom-access: no
supports-register-dump: yes
supports-priv-flags: no


  Last updated by Anagha Pamidi