In case of containerd there is no need to restart. Restart for crio is needed
when it manages the networks ns lifecycle. Removed the kubelet stop step which
was a leftover from earlier device plugin install.
Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
Addressing comments -
1a. added flag to `sriov.sh` to indicate bind to vfio-pci
1b. systemd unit sets up two pools 1 per mode
2a. updated `sriov-conf.yaml` to use the two pools
3a. included network examples for both resource pools
3b. included pod examples for both resource pools
Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
Adding a cni to mirror MAC address of VF on a veth pair with results
from ipam applied to enable DPDK apps to configure themselves when
operating sriov-network-device-plugin in vfio mode
Updated helper sriov.sh to allow for binding VFs to vfio-pci on boot
Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
Instead of relying on configmap to be created/customized by the user to
generate multus conf, we autogenerate it using the current default CNI
conf file on the host. This way CNI's with host specific conf details
will continue to work.
```
{
"type": "calico",
"nodename": "clr-01",
...
}
```
Since SR-IOV device may not be present on all nodes, the daemonset
should not exit as this will put the pod in restart loop. This would
cause the kubelet and runtimes to repeatedly restart, as per
initContainer logic.
Signed-off-by: Saikrishna Edupuganti <saikrishna.edupuganti@intel.com>
Khanak Nangia