http://bugzilla.opensuse.org/show_bug.cgi?id=1178868 Bug ID: 1178868 Summary: cilium-k8s-yaml Helm Chart does not deploys a working cilium installation Classification: openSUSE Product: openSUSE Tumbleweed Version: Current Hardware: x86-64 OS: Other Status: NEW Severity: Major Priority: P5 - None Component: Kubic Assignee: kubic-bugs@opensuse.org Reporter: contact@ffreitas.io QA Contact: qa-bugs@suse.de Found By: --- Blocker: --- Hi all, I was using kubic with cilium just fine using kubic-control builded from source (master branch). With the latest helm chart of cilium packaged in cilium-k8s-yaml it does not work anymore. By searching a bit I think there might be issues with the cilium images from registry.opensuse.org. # The problem The latest commit of kubic-control deploys cilium using the helm chert installed with the package cilium-k8s-yaml. For a while this chart was not working (some default values were missing) so a quickfix was made with custom helm values pointing to upstream images. Recently the chart updated to the version 1.8.5 of cilium and the fix is now useless. The template is now working without having to add values. To deploy the following chart without any value we need to build kubic-control from source and to deploy it on our master node. We then do the following : ``` bash echo "#We do not want values here" > /etc/kubic/helm/cilium.yaml kubicctl init --pod-network=cilium ``` But doing so, the deployed cilium is not working : ```bash node-master:~ # kubectl get pods -A NAMESPACE NAME READY STATUS RESTARTS AGE kube-system cilium-7sbtc 0/1 CrashLoopBackOff 10 30m kube-system cilium-operator-cfc74c99b-9wwhs 0/1 Pending 0 30m kube-system cilium-operator-cfc74c99b-qcqb2 0/1 ImagePullBackOff 0 30m kube-system coredns-76679c6978-dx2xc 0/1 ContainerCreating 0 30m kube-system coredns-76679c6978-pfp7c 0/1 ContainerCreating 0 30m kube-system etcd-node-master 1/1 Running 0 30m kube-system kube-apiserver-node-master 1/1 Running 0 30m kube-system kube-controller-manager-node-master 1/1 Running 0 30m kube-system kube-proxy-bzhrf 1/1 Running 0 30m kube-system kube-scheduler-node-master 1/1 Running 0 30m kube-system kured-j5dh8 0/1 ContainerCreating 0 30m ``` # cilium operator The first problem comes from the cilium-operator. The image used in the chart is registry.opensuse.org/kubic/cilium-operator-generic:1.8.5 put this image does not exist. It is due to the chart template : ```bash node-master:~ # grep gen /usr/share/k8s-helm/cilium/charts/operator/templates/deployment.yaml - cilium-operator-generic image: "{{ .Values.global.registry }}/{{ .Values.image }}-generic:{{ .Values.global.tag }}" ``` It can only be fixed by changing the chart or updating the image name in the registry. # cilium We still have the cilium pod in a CrashLoopBackOff. Looking at the logs it seems to come from the image stored in registry.opensuse.org : ```bash node-master:~ # kubectl logs -n kube-system cilium-7sbtc | tail level=info msg=" --tunnel='vxlan'" subsys=daemon level=info msg=" --version='false'" subsys=daemon level=info msg=" --write-cni-conf-when-ready=''" subsys=daemon level=info msg=" _ _ _" subsys=daemon level=info msg=" ___|_| |_|_ _ _____" subsys=daemon level=info msg="| _| | | | | | |" subsys=daemon level=info msg="|___|_|_|_|___|_|_|_|" subsys=daemon level=info msg="Cilium 1.8.5 go version go1.13.15 linux/amd64" subsys=daemon level=info msg="cilium-envoy version: 6d0a55191baac475046d13e52ffe330f3a56a4ce/1.14.4/Modified/DEBUG/BoringSSL" subsys=daemon level=fatal msg="clang: NOT OK" error="Invalid character(s) found in patch number \"0\\nTarget:\"" subsys=linux-datapath ``` Is someone already looking at this ? Is there any way for me to help with this issues ? Regards, Francisco -- You are receiving this mail because: You are the assignee for the bug.