Cilium: test: Running individual tests is no longer possible

[brb]

$ NETNEXT=true KUBEPROXY=0 K8S_NODES=3 NO_CILIUM_ON_NODE=k8s3 BUILD_NUMBER=2 ginkgo --v --focus 'K8sService.*' --focus '.*LoadBalancer.*' -- --cilium.showCommands=false --cilium.provision=false --cilium.holdEnvironment=true --cilium.testScope="k8s" --cilium.passCLIEnvironment=true
[...]
12:06:36 STEP: Installing Cilium

---
K8sServicesTest Checks service across nodes Tests NodePort BPF Tests with direct routing Tests LoadBalancer Connectivity to endpoint via LB
at /home/brb/sandbox/gopath/src/github.com/cilium/cilium/test/ginkgo-ext/scopes.go:514

[Cilium cannot be installed
Expected
    <*helpers.cmdError | 0xc0004b0e20>: Unable to generate YAML (Process exited with status 1) output: cmd: helm template --validate /home/vagrant/go/src/github.com/cilium/cilium/install/kubernetes/cilium --namespace=kube-system  --set global.autoDirectNodeRoutes=true  --set global.ci.kubeCacheMutationDetector=true  --set global.tag=latest  --set global.logSystemLoad=true  --set global.k8sServicePort=6443  --set global.tunnel=disabled  --set global.bpf.preallocateMaps=true  --set global.devices='{enp0s8,enp0s3}'  --set global.k8s.requireIPv4PodCIDR=true  --set config.sessionAffinity=false  --set hubble-relay.image.tag=latest  --set global.debug.enabled=true  --set global.hubble.listenAddress=:4244  --set global.affinity.nodeAffinity.requiredDuringSchedulingIgnoredDuringExecution.nodeSelectorTerms[0].matchExpressions[0].operator=NotIn  --set preflight.image=cilium-dev  --set global.hostFirewall=true  --set config.bpfMasquerade=true  --set global.registry=k8s1:5000/cilium  --set agent.image=cilium-dev  --set global.etcd.leaseTTL=30s  --set config.enableCnpStatusUpdates=true  --set operator.image=operator  --set global.pprof.enabled=true  --set global.affinity.nodeAffinity.requiredDuringSchedulingIgnoredDuringExecution.nodeSelectorTerms[0].matchExpressions[0].key=cilium.io/ci-node  --set global.k8sServiceHost=192.168.36.11  --set global.kubeProxyReplacement=strict  --set global.ipv4.enabled=true  --set global.hubble.enabled=true  --set global.nativeRoutingCIDR=10.0.0.0/8  --set global.ipv6.enabled=true  --set global.affinity.nodeAffinity.requiredDuringSchedulingIgnoredDuringExecution.nodeSelectorTerms[0].matchExpressions[0].values[0]=k8s3  > cilium-162acbd5129db15a.yaml
Exitcode: 1
Err: Process exited with status 1
Stdout:

Stderr:
         Error: rendered manifests contain a resource that already exists. Unable to continue with install: existing resource conflict: namespace: kube-system, name: cilium, existing_kind: /v1, Kind=ServiceAccount, new_kind: /v1, Kind=ServiceAccount

The following PR https://github.com/cilium/cilium/pull/12763/files#diff-2304891712e5d0140828499fc3a19043R2305-R2354 removed the cleanup of the resources.

Removing --validate from helm template can be used as a temporary mitigation.

[pchaigno]

For the record, the line to change to work around this bug is:
https://github.com/cilium/cilium/blob/aeefd5862ae4affe672b76803b27224f09f6343a/test/helpers/kubectl.go#L3860