Tolerate cloudprovider taints

[k0da]

Is your feature request related to a problem?

We use fleet to bootstrap CPI on cluster.

So far gitjob fails to render manifests:

kubectl get po -n fleet-local
NAME                           READY   STATUS    RESTARTS   AGE
ca-injector-921a9-65676        0/1     Pending   0          12h
cert-manager-921a9-6wqfz       0/1     Pending   0          12h
external-dns-921a9-cv2pl       0/1     Pending   0          12h
fluent-operator-921a9-j5jns    0/1     Pending   0          12h
grafana-operator-921a9-dnsqf   0/1     Pending   0          12h
ira-controller-921a9-djsfw     0/1     Pending   0          12h
p8s-921a9-4cbnj                0/1     Pending   0          12h
vsphere-csi-921a9-vhjnw        0/1     Pending   0          12h

  Warning  FailedScheduling  11m (x151 over 12h)  default-scheduler  0/2 nodes are available: 2 node(s) had untolerated taint {node.cloudprovider.kubernetes.io/uninitialized: true}. preemption: 0/2 nodes are available: 2 Preemption is not helpful for scheduling.

Solution you'd like

GitJob must tolerate common taint for cloud provider:

node.cloudprovider.kubernetes.io/uninitialized

This will allow to initially install and later maintain CPI from central place

Alternatives you've considered

No response

Anything else?

No response

[sbulage]

System Information

Rancher Version	Fleet Version
2.10.0-alpha5	fleet:v0.11.0-beta.3

Steps performed

• Created a GitRepo which deploys nginx application.
• Wait for a second to create pod.
• Navigated to Kubernetes Job created by GitJob to deploy application.
• Verified that toleration for cloud provider uninitialized is present on it. (See below screenshot for details.)

Note: In latest version of Fleet jobs are removed immediately, I opened Rancher in 2 separate tabs to open jobs YAML before it get removed.

Taints on GitJob's Job