Hi all, I’m trying to upgrade the version of rke on our clusters. We have openstack as the cloud provisioner with rke 1.2.7. If we try to create a new cluster using that with the in-tree openstack provider, it all works fine. I have been trying to upgrade to 1.3.12 (also tried 1.3.14), when I try that version with the in-tree openstack provider the connection to the openstack API fails. I have been trying to use the external cloud provisioner and apply the helm chart for the openstack cloud controller manager, but that also fails with a message about

failed to try resolving symlinks in path…

. I can’t find any information in the rke documentation about using the external cloud provider, so I’m not sure if it is supported (I may have missed it!). Has anyone successfully used rke with the external cloud provider with the Openstack cloud controller manger and the cinder CSI plugin?

has anyone configured rke with the GracefulNodeShutdown feature? It's on by default in the latest versions of k8s, but I'm not sure where to set the ShutdownGracePeriod and ShutdownGracePeriodCriticalPods parameters to the kubelet

I write here for error and I can't delete the message. Sorry.

Installing RKE2 agent on a windows worker. Should I just use the URL given from ranchers custom clusterr -> registration. Or should I follow instructions from https://docs.rke2.io/install/windows_airgap/ (I cant find any other installation documentation for the agent)

I get an error while deploying RKE with 1 master(role:controlplane,etcd) and 2 workers(role: worker). Following is my cluster.yml

nodes: - address: <<VM-1>> port: "22" internal_address: "" role: - controlplane - etcd hostname_override: "" user: vudemaster docker_socket: /var/run/docker.sock ssh_key: "" ssh_key_path: ~/.ssh/id_rsa ssh_cert: "" ssh_cert_path: "" labels: {} taints: [] - address: <<VM-2>> port: "22" internal_address: "" role: - worker hostname_override: "" user: vudeadmin docker_socket: /var/run/docker.sock ssh_key: "" ssh_key_path: ~/.ssh/id_rsa ssh_cert: "" ssh_cert_path: "" labels: {} taints: [] - address: <<VM-3>> port: "22" internal_address: "" role: - worker hostname_override: "" user: vudeadmin docker_socket: /var/run/docker.sock ssh_key: "" ssh_key_path: ~/.ssh/id_rsa ssh_cert: "" ssh_cert_path: "" labels: {} taints: [] Error: FATA[0115] [workerPlane] Failed to bring up Worker Plane: [Failed to verify healthcheck: Failed to check http://localhost:10256/healthz for service [kube-proxy] on host [cdedops-vude4.elements.local]: Get "http://localhost:10256/healthz": Unable to access the service on localhost:10256. The service might be still starting up. Error: ssh: rejected: connect failed (Connection refused), log: E0919 093712.439028 2897 node.go:161] Failed to retrieve node info: Get "https://127.0.0.1:6443/api/v1/nodes/cdedops-vude4.elements.local": read tcp 127.0.0.148478 &gt;127.0.0.16443: read: connection reset by peer] . But If I would specify roles as [controlplane,etcd] in all nodes, then it works fine. The error comes when role is given as worker alone. Refer configs above. Please suggest.

Note :: W.r.t above error, all the required pre-requisites for cluster enablement such as SSH configs, dockers has already been done on all nodes. I am not using Rancher UI, I am using plain RKE only. rke version v1.3.11.

I can not found the list of required kernel modules for RKE installation

Hi , I have a issue which is I make deployment to rancher, code base is laravel. my application run for 6 hrs or 5 hrs and its stop working on rancher. when i redeploy application its start working again what should be the error. i have one pod running against application. Please help Thanks

Hi guys! Is it still not possible to use Calico with RKE for bare metal? Here it states that we only can use it for aws or gce https://rancher.com/docs/rke/latest/en/config-options/add-ons/network-plugins/#calico

Hi everyone. I’m trying ti upgrade ingress version from

cluster.yml

. I’m adding image to

system_images

block in

cluster.yml

.After saving it my changes does’t getting saved and in ingress the same old image is used,

Hi community, I’m trying to expand Rancher RKE1 running as network plugin Calico. The current cluster is formed by three Lenovo SR630 servers, Now I want to add three Virtual machines as control plane and etcd nodes, but when I make the expansion I notice an error on readiness probe of calico with following message: “Readiness probe failed: calico/node is not ready: BIRD is not ready: Error querying BIRD: unable to connect to BIRDv4 socket: dial unix /var/run/bird/bird.ctl: connect: no such file or directory”. On the logs of the calico node I also noticed the following message: “bird: Netlink: File exists” I searched possible solutions but I don’t find any solutions. I also recreated the same network scheme on the vms, creating linux bond as the one I have on the Lenovo servers. There is anybody on this channel that faced similar problems?

Hi, what cgroup-driver should I set for rke1, cgroupfs, is this only apply to kubelet using systemd (which is not rke case )?

I have RKE clusters managed by Rancher running on top of Harvester. I am unable to provision the cluster to add new etcd members because RKE is currently attempting to remove members during reconciliation which do not currently exist — a process which fails or hangs (can’t tell which)

I have a production cluster with virtual machines for the control plane and bare metal servers for workers. When I run rke up/update, I add extra args to my cluster.yaml for kubelet to respect certain reserved memory and cpu On multipurpose nodes. However only the bare metal nodes need this and if it applies to master nodes, then they will use the reserved cpu and memory amounts which I don’t want (only want that for workers). Is there a way to target specific nodes for the cluster.yaml to apply to? This is blocking me from upgrading kubernetes.

Hi, I am creating new RKE cluster using Rancher UI with the help of VMware vSphere provider. After creating RKE cluster, it created new VMs on vSphere and then connect it and ... Finally, the cluster status in Rancher is shown "Waiting for API to be available" and the cluster is not usable. When I check the cluster condition, it is shown like this. can you share your opinion to fix this issue?

77

@creamy-pencil-82913 Looks like I need your help again! Looks like the rancher provisioned nodes don't request DHCP lease renewal. After a reboot of the nodes, their IPs changed and broke the cluster. I tried to scale down to single node and run

rke2 server --cluster-reset

but due to an accident, rancher marked all nodes for deletion. I quickly exported an etcd snapshot of the database because rancher is now creating brand new nodes with a new database. I also disabled the account the vSphere Cloud Provider uses to talk to vcenter to stop the new cluster from deleting the volumes. It happened to me before. Is there a way in which I can safely restore the etcd database on the newly created nodes?

Hello guys. I am trying to find some info about migration from rke1 to rke2

Any help?

I mean official migration blessed by rancher

Help me. after "rke up" i cannot login rancher, i get pod and check log almost relate to certificate "x509: certificate is valid for 10.32.0.81, 10.32.0.82, 10.32.0.83, 127.0.0.1, 10.43.0.1, "

@boundless-smartphone-66270 Please follow the documentation to save your kube config file: https://rancher.com/docs/rke/latest/en/installation/#deploying-kubernetes-with-rke

Hi, I currently have a cluster in rancher with

rancher_kubernetes_engine_config: 
  network
    mtu: 0
    options:
      flannel_backend_type: vxlan
    plugin: flannel

when I change it to:

rancher_kubernetes_engine_config:
  network:
    plugin: none

will it remove the network plugin? I want to install Calico manually and wondering if that would work.

Hey folks. If I lost my rkestate and rancher_cluster.yml files, is there anyway to regenerate them from an active cluster?

hey folks, i recently deployed an rke cluster on baremetal servers running oracle linux 7. After deployment and i deploy a pod i get the following error for all outgoing https traffic

14094410:SSL routines:ssl3_read_bytes:sslv3 alert handshake failure

see openssl response from pod

bash-5.1# openssl s_client -showcerts -connect <http://accounts.google.com:443|accounts.google.com:443>

CONNECTED(00000003)

139739654130504:error:14094410:SSL routines:ssl3_read_bytes:sslv3 alert handshake failure:ssl/record/rec_layer_s3.c:1543:SSL alert number 40

---

no peer certificate available

---

No client certificate CA names sent

---

SSL handshake has read 7 bytes and written 321 bytes

Verification: OK

---

New, (NONE), Cipher is (NONE)

Secure Renegotiation IS NOT supported

No ALPN negotiated

Early data was not sent

Verify return code: 0 (ok)

---

i’ve been having some issues with RKE 1.3 and clusters made of EC2 instances in AWS. With RKE 1.2 I can run

rke up

repeatedly with no issues. Using the same cluster state file and RKE 1.3, I continually get stuck with one of the nodes not able to be found. I can’t find anything online that is an exact match for this problem. Anyone else experience this? I opened this github issue with details https://github.com/rancher/rke/issues/3091

Is RKE 1.4.0 a mistake release? What’s different from 1.3.16?

Hi Team I am unable to create Vmware cloud credentials in Rancher server. UI is throwing below error message. Can anyone help in this regard. Index with name by-cloud-cred does not exist