I want to know that how to use api to curd my crd resources with api. And I can write a sdk to control resources.
Use kubectl just
kubectl get inferenceservices test-sklearn -n kserve-test
kubectl apply -f xx.yaml -n kserve-test
kubectl delete -f xx.yaml -n kserve-test
apiVersion: "serving.kubeflow.org/v1beta1"
kind: "InferenceService"
metadata:
name: "test-sklearn"
spec:
predictor:
sklearn:
storageUri: "http://xxxx"
get call process in process_log
kubectl get inferenceservices test-sklearn -n kserve-test --v=8 > process_log 2>&1
use kubectl proxy to test
kubectl proxy --address 0.0.0.0 --accept-hosts=^.*
TEST to get resource status
GET http://xxx:8001/apis/serving.kubeflow.org/v1beta1/namespaces/kserve-test/inferenceservices/test-sklearn
Related
I want to create service account for native Kubernetes cluster so that I can send API calls:
kubernetes#kubernetes1:~$ kubectl create serviceaccount user1
serviceaccount/user1 created
kubernetes#kubernetes1:~$ kubectl create clusterrole nodeaccessrole --verb=get --verb=list --verb=watch --resource=nodes
clusterrole.rbac.authorization.k8s.io/nodeaccessrole created
kubernetes#kubernetes1:~$ kubectl create clusterrolebinding nodeaccessrolebinding --serviceaccount=default:user1 --clusterrole=nodeaccessrole
clusterrolebinding.rbac.authorization.k8s.io/nodeaccessrolebinding created
kubernetes#kubernetes1:~$
kubernetes#kubernetes1:~$ kubectl get serviceaccount user1
NAME SECRETS AGE
user1 0 7m15s
kubernetes#kubernetes1:~$
Do you know how I can get the token?
SOLUTION for v1.25.1:
kubectl create sa cicd
kubectl get sa,secret
cat <<EOF | kubectl apply -f -
apiVersion: v1
kind: Pod
metadata:
name: cicd
spec:
serviceAccount: cicd
containers:
- image: nginx
name: cicd
EOF
kubectl exec cicd -- cat /run/secrets/kubernetes.io/serviceaccount/token && echo
kubectl exec cicd cat /run/secrets/kubernetes.io/serviceaccount/token && echo
kubectl create token cicd
kubectl create token cicd --duration=999999h
cat <<EOF | kubectl apply -f -
apiVersion: v1
kind: Secret
type: kubernetes.io/service-account-token
metadata:
name: cicd
annotations:
kubernetes.io/service-account.name: "cicd"
EOF
kubectl get sa,secret
kubectl describe secret cicd
kubectl describe sa cicd
kubectl get sa cicd -oyaml
kubectl get sa,secret
One thing is not clear:
kubectl exec cicd -- cat /run/secrets/kubernetes.io/serviceaccount/token && echo
kubectl exec cicd cat /run/secrets/kubernetes.io/serviceaccount/token && echo
should I use '--' into the above commands?
If you just want to retrieve the token from the given SA you can simply execute:
kubectl get secret $(kubectl get sa <sa-name> -o jsonpath='{.secrets[0].name}' -n <namespace>) -o jsonpath='{.data.token}' -n <namespace> | base64 --decode
Feel free to remove the | base64 --decode if you don't want to decode. Just as a side node, this command might need to be amended depending on the type of secret, however for your use-case this should work
Once you have your value you can execute curl commands, such as:
curl -k -H "Authorization: Bearer $TOKEN" -X GET "https://<KUBE-API-IP>:6443/api/v1/nodes"
I am trying to set up Argo CD on Google Kubernetes Engine Autopilot and each pod/container is defaulting to the default resource request (0.5 vCPU and 2 GB RAM per container). This is way more than the pods need and is going to be too expensive (13GB of memory reserved in my cluster just for Argo CD). I am following the Getting Started guide for Argo CD and am running the following command to add Argo CD to my cluster:
kubectl apply -n argocd -f https://raw.githubusercontent.com/argoproj/argo-cd/stable/manifests/install.yaml
How do I specify the resources for each pod when I am using someone else's yaml template? The only way I have found to set resource requests is with my own yaml file like this:
apiVersion: v1
kind: Pod
metadata:
name: memory-demo
namespace: mem-example
spec:
containers:
- name: memory-demo-ctr
image: polinux/stress
resources:
limits:
memory: "200Mi"
requests:
memory: "100Mi"
But I don't understand how to apply this type of configuration to Argo CD.
Thanks!
So right now you are just using kubectl with the manifest from github and you cannot edit it. What you need to do is
1 Download the file with wget
https://raw.githubusercontent.com/argoproj/argo-cd/stable/manifests/install.yaml
2 Use an editor like nano or vim to edit the file with requests as
explained in my comments above.
3 Then use kubectl apply -f newfile.yaml
You can dump the yaml of argocd, then customize your resource request, and then apply the modified yaml.
$ kubectl get deployment -n argocd -o yaml > argocd_deployment.yaml
$ kubectl get sts -n argocd -o yaml > argocd_statefulset.yaml
$ # modify resource
$ vim argocd_deployment.yaml
$ vim argocd_statefulset.yaml
$ kubectl apply -f argocd_deployment.yaml
$ kubectl apply -f argocd_statefulset.yaml
Or modify deplopyment and statefulset directly by kubectl edit
$ kubectl edit deployment -n argocd
$ kubectl edit sts -n argocd
It seems that --dry-run flag is not available for service.
kubectl create service --
--add-dir-header --log-backtrace-at --server
--alsologtostderr --log-dir --skip-headers
--as --log-file --skip-log-headers
--as-group --log-file-max-size --stderrthreshold
--cache-dir --log-flush-frequency --tls-server-name
--certificate-authority --logtostderr --token
--client-certificate --match-server-version --user
--client-key --namespace --username
--cluster --password --v
--context --profile --vmodule
--insecure-skip-tls-verify --profile-output --warnings-as-errors
--kubeconfig --request-timeout
Is there a way to create a service yaml file without --dry-run=client option. I tried with the below command and getting an error.
kubectl create service ns-service nodeport --dry-run=client -o yaml >nodeport.yaml
Error: unknown flag: --dry-run
See 'kubectl create service --help' for usage.
There are two ways to do this.
=================================================================
First Way:- using kubectl create service
What wrong you are doing here is you are giving service name befor the service type in command that's why its failing.
correct way is
Syntax :
kubectl create service clusterip NAME [--tcp=<port>:<targetPort>] [--dry-run=server|client|none] [options]
Example :
kubectl create service nodeport ns-service --tcp=80:80 --dry-run=client -o yaml
=================================================================
Second way:-
Here you can use kubectl expose command to create a service file.
Let's assume you have a pod running with the name nginx. and you want to create a service for nginx pod.
then I will write below command to generate the service file.
Synatax:
kubectl expose [pod/deployment/replicaset] [name-of-pod/deployment/replicaset] --port=80 --target-port=8000 --dry-run=client -o yaml
Example:
kubectl expose pod nginx --port=80 --target-port=8000 --dry-run=client -o yaml
output:
apiVersion: v1
kind: Service
metadata:
creationTimestamp: null
labels:
run: nginx
name: nginx
spec:
ports:
- port: 80
protocol: TCP
targetPort: 8000
selector:
run: nginx
status:
loadBalancer: {}
I am using kubectl apply -f https://k8s.io/examples/controllers/nginx-deployment.yaml to create deployment.
I want to create deployment in my namespace examplenamespace.
How can I do this?
There are three possible solutions.
Specify namespace in the kubectl command:
kubectl apply -f https://k8s.io/examples/controllers/nginx-deployment.yaml -n my-namespace
Specify namespace in your yaml files:
apiVersion: apps/v1
kind: Deployment
metadata:
name: my-deployment
namespace: my-namespace
Change default namespace in ~/.kube/config:
apiVersion: v1
kind: Config
clusters:
- name: "k8s-dev-cluster-01"
cluster:
server: "https://example.com/k8s/clusters/abc"
namespace: "my-namespace"
By adding -n namespace to command you already have. It also works with other types of resources.
kubectl apply -f https://k8s.io/examples/controllers/nginx-deployment.yaml -n namespacename
First you need to create the namespace likes this
kubectl create ns nameOfYourNamespace
Then you create your deployment under your namespace
kubectl apply -f https://k8s.io/examples/controllers/nginx-deployment.yaml -n examplenamespace
The ns at
kubectl create ns nameOfYourNamespace
stands for namespace
The -n
kubectl apply -f https://k8s.io/examples/controllers/nginx-deployment.yaml -n examplenamespace
stands for --namespace
So you first create your namespace in order Kubernetes know what namespaces dealing with.
Then when you are about to apply your changes you add the -n flag that stands for --namespace so Kubernetes know under what namespace will deploy/ create the proper resources
I am trying to define the namespace name when executing the kubectl create deployment command?
This is what I tried:
kubectl create deployment test --image=banu/image1 namespace=test
and this doesn't work.
And I want to expose this deployment using a ClusterIP service within the cluster itself for that given namespace How can I do that using kubectl command line?
You can specify either -n or --namespace options.
kubectl create deployment test --image=nginx --namespace default --dry-run -o yaml and see result deployment yaml.
Using kubectl run
kubectl run test --namespace test --image nginx --port 9090 --dry-run -o yaml
You need to create a namespace like this
kubectl create ns test
ns stands for namespace, so with kubectl you say you want to create namespace with name test
Then while you creating the deployment you add the namespace you want
kubectl create deployment test --image=banu/image1 -n test
Where flag -n stands for namespace, that way you say to Kubernetes that all resources related to that deployment will be under the test namespace
In order to see all the resources under a specific namespace
kubectl get all -n test
--namespace and -n is the same things
Use -n test instead of namespace=test
Sample with nginx image:
$ kubectl create deployment nginx --image=nginx -n test
deployment.apps/nginx created
$ kubectl get deploy -n test
NAME READY UP-TO-DATE AVAILABLE AGE
nginx 1/1 1 1 8s
In second case you need to create service and define labels from deployment.
You can find correct labels by runnig something like:
kubectl -n test describe deploy test |grep Labels:
and apply service like:
apiVersion: v1
kind: Service
metadata:
name: test-svc
namespace: test
spec:
ports:
- name: test
port: 80 # Change this port
protocol: TCP
type: ClusterIP
selector:
# Here you need to define output from previous step