Friday, 2021-07-09

kata-irc-bot<vasek.mlejnsky> > Curious how you were able to enable virt? GKE doesn't allow to set a custom boot image for nodes but you can do a little workaround and edit the existing image. The solution is very GCP specific: You have to edit a template that GCP is using to deploy it's nodes (= compute engine VMs). You enable the virt, in the new template and replace the old template with the new one. The cluster then restarts itself and deploy nodes06:36
kata-irc-botbased on the new template. It's based on this solution - https://issuetracker.google.com/issues/110507927#comment22  > Did you apply rbac? Yes. Once I had virt enabled, I ran these to install Kata: ```# Kata kubectl apply -f https://raw.githubusercontent.com/kata-containers/kata-containers/main/tools/packaging/kata-deploy/kata-rbac/base/kata-rbac.yaml kubectl apply -f06:36
kata-irc-bothttps://raw.githubusercontent.com/kata-containers/kata-containers/main/tools/packaging/kata-deploy/kata-deploy/base/kata-deploy.yaml  # Runtime classes kubectl apply -f https://raw.githubusercontent.com/kata-containers/kata-containers/main/tools/packaging/kata-deploy/runtimeclasses/kata-runtimeClasses.yaml  # Example FC workload kubectl apply -f06:36
kata-irc-bothttps://raw.githubusercontent.com/kata-containers/kata-containers/main/tools/packaging/kata-deploy/examples/test-deploy-kata-fc.yaml``` > What do you see in the logs of the kata-deploy daemonset?  The daemonset doesn't output any logs because its pod never gets fully ready. Here's what I see when I run `kubectl logs -f daemonset/kata-deploy -n="kube-system"` : ```Error from server (BadRequest): container "kube-kata" in pod "kata-deploy-5cxwv"06:36
kata-irc-botis waiting to start: ContainerCreating``` I attached the output of `kubectl describe daemonset kata-deploy -n="kube-system"` in the `kata-deploy-daemonset.txt` file.  I also attached the output of `kubectl describe pod kata-deploy-5cxwv -n=kube-system` for the daemonset's pod in the `kata-deploy-daemonset-pod.txt` file.  > Also, can you share `kubectl get nodes --show-labels`?  I attached the command's outputs before Kata is installed and06:36
kata-irc-botafter I run the `kubectl apply` commands mentioned above in the `get-node-labels-before-kata-install.txt` file and the `get-node-labels-after-kata-install.txt` file06:36
kata-irc-bot<sidorov.ml99> Hi everyone! I’m trying to use kata with k3s, have deployed kata with kata-deploy and created a test deployment with kata-fc runtime. The pod is stuck in ContainerCreating status and posts this events: ```Events:   Type     Reason                  Age   From               Message   ----     ------                  ----  ----               -------   Normal   Scheduled               30s   default-scheduler  Successfully assigned11:57
kata-irc-botdefault/php-apache-kata-fc-5ccb8df89-bxdr2 to k3s-master-1   Warning  FailedCreatePodSandBox  28s   kubelet            Failed to create pod sandbox: rpc error: code = Unknown desc = failed to create containerd task: failed to mount "/run/kata-containers/shared/containers/e395498e3b0ac41dcda036c05eb13001f4ccba99037161c865964d0e2b3f6657/rootfs" to "/run/kata-containers/e395498e3b0ac41dcda036c05eb13001f4ccba99037161c865964d0e2b3f6657/rootfs",11:57
kata-irc-botwith error: No such file or directory (os error 2): unknown   Warning  FailedCreatePodSandBox  13s   kubelet            Failed to create pod sandbox: rpc error: code = Unknown desc = failed to create containerd task: failed to mount "/run/kata-containers/shared/containers/a6193715cd634172aeb4016771e28d39461ead562846e94fdf22285bc67718e1/rootfs" to "/run/kata-containers/a6193715cd634172aeb4016771e28d39461ead562846e94fdf22285bc67718e1/rootfs",11:57
kata-irc-botwith error: No such file or directory (os error 2): unknown``` Any ideas what could cause this issue? I used to have similar problem when running kata with cri-o with overlayfs storage, maybe this is devicemapper related11:57
kata-irc-bot<sidorov.ml99> Actually firecracker is the only runtime that doesn’t seem to work: `NAME                                    READY   STATUS              RESTARTS   AGE` `php-apache-kata-clh-796cd8db47-d6bzg    1/1     Running             0          4m11s` `php-apache-kata-fc-5ccb8df89-bxdr2      0/1     ContainerCreating   0          19m` `php-apache-kata-qemu-6bbd4d8c75-frtnp   1/1     Running             0          3m58s`12:05
kata-irc-bot`php-apache-runc-5cd4cc5f96-lf7kw        1/1     Running             0          3m48s`12:05
kata-irc-bot<salvador.fuentes> Hi, yes, FC requires to have devicemapper. Is this with containerd? could you paste your containerd config file? and could you confirm that you are using devicemapper?12:55
kata-irc-bot<anastassios.nanos> hi there! regarding storage, we haven't tried anything else than devicemapper. I think its the only choice for now.  regarding >1 container per pod, we haven't tried that either -- I remember looking at an issue a couple of months back about that but can't find it right now. Have you tried to qemu equivalent?14:57
kata-irc-bot<anastassios.nanos> yeap, again, we have only tried with devicemapper. I think it's the only solution at the moment.14:58
kata-irc-bot<sidorov.ml99> Ok, it seems like containerd for k3s wasn’t compiled with devicemapper plugin support, so it’s not possible to run kata-fc16:03
kata-irc-bot<sidorov.ml99> I have created thin pool and changed the config but it doesn’t change anything16:04
kata-irc-bot<lagkasorestis> Hello, I have tried running k3s with fc with @anastassios.nanos.. we are actually running it now on a rasperi pi4.. k3s has a lightweight version of containerd which does not include the devicemapper plugin if you run `k3s ctr plugin ls`  you will see that there is no devmapper16:06
kata-irc-bot<lagkasorestis> so there are 2 solutions: • either recompile k3s containerd and include the devmapper plugin • or run k3s with external containerd...16:07
kata-irc-bot<sidorov.ml99> yep, seems like that, thanks!16:08
kata-irc-bot<lagkasorestis> btw if you already have configured the host's containerd with devmapper the the following should install k3s to run with your systems containerd `curl -sfL https://get.k3s.io | INSTALL_K3S_EXEC="--container-runtime-endpoint unix:///run/containerd/containerd.sock" sh -`16:10

Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!