I've a EKS setup (v1.16) with 2 ASG: one for compute ("c5.9xlarge") and the other gpu ("p3.2xlarge").
Both are configured as Spot and set with desiredCapacity 0.
K8S CA works as expected and scale out each ASG when necessary, the issue is that the newly created gpu instance is not recognized by the master and running kubectl get nodes emits nothing.
I can see that the ec2 instance was in Running state and also I could ssh the machine.
I double checked the the labels and tags and compared them to the "compute".
Both are configured almost similarly, the only difference is that the gpu nodegroup has few additional tags.
Since I'm using eksctl tool (v.0.35.0) and the compute nodeGroup vs. gpu nodeGroup is basically copy&paste, I can't figured out what could be the problem.
UPDATE:
ssh the instance I could see the following error (/var/log/messages)
failed to run Kubelet: misconfiguration: kubelet cgroup driver: "systemd" is different from docker cgroup driver: "cgroupfs"
and the kubelet service crashed.
would it possible the my GPU uses wrong AMI (amazon-eks-gpu-node-1.18-v20201211)?
As a simple you can use this preBootstrapCommands in eksctl yaml config file:
- name: test-node-group
preBootstrapCommands:
- "sed -i 's/cgroupDriver:.*/cgroupDriver: cgroupfs/' /etc/eksctl/kubelet.yaml"
There is some issue with EKS 1.16, even the graviton processors machine won't join the cluster. To fix it first you try upgrading your CNI version. Please refer the documentation here:
https://docs.aws.amazon.com/eks/latest/userguide/cni-upgrades.html
And if that doesn't work, then upgrade your EKS version to the latest available version then should work.
I've found out the issue. It seems to be mis-alignment between eksctl (v0.35.0) and the AL2-GPU AMI.
AWS team change the control group in docker to be "systemd" instead of "cgroup" (github) while the eksctl tool I used didn't absorb the changes.
A temporary solution is to edit the /etc/eksctl/kubelet.yaml file using preBootstrapCommands
Related
We’re providing our own AMI node images for EKS using the self-managed node feature.
The challenge I’m currently having is how to fetch the kubernetes version from within the EKS node as it starts up.
I’ve tried IMDS - which unfortunately doesn’t seem to have it:
root#ip-10-5-158-249:/# curl -H "X-aws-ec2-metadata-token: $TOKEN" http://169.254.169.254/latest/meta-data/
ami-id
ami-launch-index
ami-manifest-path
autoscaling/
block-device-mapping/
events/
hostname
iam/
identity-credentials/
instance-action
instance-id
instance-life-cycle
instance-type
local-hostname
local-ipv4
mac
metrics/
network/
placement/
profile
reservation-id
It also doesn’t seem to be passed in by the EKS bootstrap script - seems AWS is baking a single K8s version into each AMI. (install-worker.sh).
This is different from Azure’s behaviour of baking a bunch of Kubelets into a single VHD.
I’m hoping for something like IMDS or a passed in user-data param which can be used at runtime to symlink kubelet to the correct kubelet version binary.
Assumed you build your AMI base on the EKS optimized AMI; one of the possible way is use kubelet --version to capture the K8s version in your custom built; as you knew EKS AMI is coupled with the control plane version. If you are not using EKS AMI, you will need to make aws eks describe-cluster call to get cluster information in order to join the cluster; which the version is provided at cluster.version.
I am trying to reproduce an issue that requires me to use containerd v1.4.4 for my container-runtime and kubernetes v1.19.8. When I try to use minikube to create a multi-node cluster locally, it allows me to specify the kubernetes version but I am unable to specify the containerd version(i.e. it always uses v1.4.9) and based on this github discussion, it doesn't seem to support it. I then turned to kind but was unable to find a way to specify the same from the documentation. Is there a way either in kind or in minikube to specify the containerd version?
I ended up using kubeadm and set up a master and worker node using 2 VMs. This allowed me to specify the versions I want on the worker node. Building a base image on kind should also work as user Mikolaj.S mentioned
Hope someone can help me.
To describe the situation in short, I have a self managed k8s cluster, running on 3 machines (1 master, 2 worker nodes). In order to make it HA, I attempted to add a second master to the cluster.
After some failed attempts, I found out that I needed to add controlPlaneEndpoint configuration to kubeadm-config config map. So I did, with masternodeHostname:6443.
I generated the certificate and join command for the second master, and after running it on the second master machine, it failed with
error execution phase control-plane-join/etcd: error creating local etcd static pod manifest file: timeout waiting for etcd cluster to be available
Checking the first master now, I get connection refused for the IP on port 6443. So I cannot run any kubectl commands.
Tried recreating the .kube folder, with all the config copied there, no luck.
Restarted kubelet, docker.
The containers running on the cluster seem ok, but I am locked out of any cluster configuration (dashboard is down, kubectl commands not working).
Is there any way I make it work again? Not losing any of the configuration or the deployments already present?
Thanks! Sorry if it’s a noob question.
Cluster information:
Kubernetes version: 1.15.3
Cloud being used: (put bare-metal if not on a public cloud) bare-metal
Installation method: kubeadm
Host OS: RHEL 7
CNI and version: weave 0.3.0
CRI and version: containerd 1.2.6
This is an old, known problem with Kubernetes 1.15 [1,2].
It is caused by short etcd timeout period. As far as I'm aware it is a hard coded value in source, and cannot be changed (feature request to make it configurable is open for version 1.22).
Your best bet would be to upgrade to a newer version, and recreate your cluster.
we are running multiple kubespray deployed clusters with 10-100 nodes.
with 1.20 kubernetes deperecates dockershim support -> https://github.com/kubernetes/kubernetes/blob/ab32085bf36fc7af1ded30456e2f09399dc1115f/CHANGELOG/CHANGELOG-1.20.md#deprecation
how to change the container runtime to containerd - without removing nodes and without destroying master.
i am not at panick, just wan't to be prepared we are at 1.19 already so 1.22 is not soo faar away.
anyways i tested it with a smaller cluster, and it was way easier as expected.
change: container_manager to containerd.
run the kubespray cluster.yml playbook over all nodes and boom.
only needed to do a simple ansible playbook to uninstall docker et-all, but it also works with docker installed.
Please treat this answer as a friendly advise.
First of all, as suggested in yesterday's fresh article Don't Panic: Kubernetes and Docker:
You do not need to panic :)
Kubernetes is only deprecating Docker as a container runtime after v1.20. They are currently only planning to remove Docker runtime support in the 1.22 release in late 2021(almost year!), so please don't brake your 100 nodes clusters till work solution will appear :)
Once we increase load by using JMeter client than my deployed service is interrupted and on GCP/GKE console it says that -
Upgrading cluster master
The values shown below are going to change soon.
And my kubectl client throw this error during upgrade -
Unable to connect to the server: dial tcp 35.236.238.66:443: connectex: No connection could be made because the target machine actively refused it.
How can I stop this upgrade or prevent my service interruption ? If service will be intrupted than there is no benefit of this auto scaling. I am new to GKE, please let me know if I am missing any configuration or parameter here.
I am using this command to create my cluster-
gcloud container clusters create ajeet-gke --zone us-east4-b --node-locations us-east4-b --machine-type n1-standard-8 --num-nodes 1 --enable-autoscaling --min-nodes 4 --max-nodes 16
It is not upgrading k8s version. Because it works fine with smaller load but as I increase load than cluster starts upgrade of master. So it looks the master is resizing itself for more nodes. After upgrade I can see more nodes on GCP console. https://github.com/terraform-providers/terraform-provider-google/issues/3385
Below command says auto scaling is not enabled on instance group.
> gcloud compute instance-groups managed list
NAME AUTOSCALED LOCATION SCOPE ---
ajeet-gke-cluster- no us-east4-b zone ---
default-pool-4***0
Workaround
Sorry forget to update it here, I found a workaround to fix it - after splitting cluster creation command in to two steps cluster is auto scaling without restarting master node:
gcloud container clusters create ajeet-ggs --zone us-east4-b --node-locations us-east4-b --machine-type n1-standard-8 --num-nodes 1
gcloud container clusters update ajeet-ggs --enable-autoscaling --min-nodes 1 --max-nodes 10 --zone us-east4-b --node-pool default-pool
To prevent this you should always create your cluster with hardcoded cluster version to the last version available.
See the documentation: https://cloud.google.com/kubernetes-engine/docs/concepts/cluster-architecture#master
This means that Goolge is managing the master, meaning that if your master is not up to date it will be updated to be in the last version and allow google to limit the number of version currently managed. https://cloud.google.com/kubernetes-engine/docs/concepts/regional-clusters
Now why do you have an interruption of service during the update: because you are in zonal mode with only one master, to prevent this you should go in regional cluster mode with more than one master, allowing for clean rolling update.
The master won't resize the node, unless the autoscaling feature is enabled in it.
As mentioned in above answer, this is a feature at the node-pool level. By looking at description of the issue, it does seems like 'autoscaling' is enabled on your node-pool and eventually a GKE's cluster autoscaler automatically resizes clusters based on the demands of the workloads you want to run(ie when there are pods that are not able to be scheduled due to resource shortages such as CPU).
Additionaly, Kubernetes cluster autoscaling does not use the Managed Instance Group autoscaler. It runs a cluster-autoscaler controller on the Kubernetes master that uses Kubernetes-specific signals to scale your nodes.
It is therefore, highly recommended not use(or rely on the autoscaling status showed by MIG) Compute Engine's autoscaling feature on instance groups created by Kubernetes Engine.