I have created the following ReplicaSet
apiVersion: extensions/v1beta1
kind: ReplicaSet
metadata:
name: nginx-test
spec:
replicas: 2
template:
metadata:
name: nginx
namespace: default
labels:
env: beta
spec:
containers:
- name: nginx
image: nginx
imagePullPolicy: IfNotPresent
ports:
- containerPort: 80
nodeSelector:
domain: cloud
This runs the pods in a node marked as cloud. Now I change the nodeSelector using the command
kubectl edit rs/nginx-test
And change the nodeSelector to edge. However the pods are not moved to the edge node. This is working for a Deployment, however not for ReplicaSet. Any ideas
Here are my 2 nodes:
NAME STATUS AGE VERSION LABELS
x1 Ready 5d v1.8.4 beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,domain=cloud,kubernetes.io/hostname=xxxx,node-role.kubernetes.io/master=
x2 Ready 5d v1.8.3 beta.kubernetes.io/arch=amd64,beta.kubernetes.io/os=linux,domain=edge,kubernetes.io/hostname=xxxx
Official Kubernetes documentation recommends that you use a Deployment, which creates ReplicaSets, rather than use ReplicaSets directly.
"A ReplicaSet ensures that a specified number of pod replicas are running at any given time. However, a Deployment is a higher-level concept that manages ReplicaSets and provides declarative updates to pods along with a lot of other useful features. Therefore, we recommend using Deployments instead of directly using ReplicaSets, unless you require custom update orchestration or don’t require updates at all."
https://kubernetes.io/docs/concepts/workloads/controllers/replicaset/
It is not unheard of to use a ReplicaSet by itself, but generally not recommended.
If a Deployment works for you, I recommend sticking with that, unless you are using ReplicaSets for some custom thing that Deployments don't work for.
Related
I create a yaml file to create rabbitmq kubernetes cluster. I can see pods. But when I write kubectl get deployment. I cant see there. I can't access to rabbitmq ui page.
apiVersion: v1
kind: Service
metadata:
labels:
app: rabbit
name: rabbit
spec:
ports:
- port: 5672
protocol: TCP
name: mqtt
- port: 15672
protocol: TCP
name: ui
type: NodePort
---
apiVersion: apps/v1
kind: StatefulSet
metadata:
name: rabbit
spec:
serviceName: rabbit
replicas: 3
selector:
matchLabels:
app: rabbit
template:
metadata:
labels:
app: rabbit
spec:
containers:
- name: rabbitmq
image: rabbitmq
nodeSelector:
rabbitmq: "clustered"
#arghya-sadhu's answer is correct.
NB I'm unfamiliar with RabbitMQ but you may need to use a different image (see 'Management Plugin`) to include the UI.
See below for more details.
You should be able to hack your way to the UI on one (!) of the Pods via:
PORT=8888
kubectl port-forward pod/rabbit-0 --namespace=${NAMESPACE} ${PORT}:15672
And then browse localhost:${PORT} (if 8888 is unavailable, try another).
I suspect (!) this won't work unless you use the image with the management plugin.
Plus
The Service needs to select the StatefulSet's Pods
Within the Service spec you should add perhaps:
selector:
app: rabbit
Presumably (!?) you are using a private repo (because you have imagePullSecrets).
If you don't and wish to use DockerHub, you may remove the imagePullSecrets section.
It's useful to document (!) container ports albeit not mandatory:
In the StatefulSet
ports:
- containerPort: 5672
- containerPort: 15672
Debug
NAMESPACE="default" # Or ...
Ensure the StatefulSet is created:
kubectl get statesfulset/rabbit --namespace=${NAMESPACE}
Check the Pods:
kubectl get pods --selector=app=rabbit --namespace=${NAMESPACE}
You can check the the Pods are bound to a (!) Service:
kubectl describe endpoints/rabbit --namespace=${NAMESPACE}
NB You should see 3 addresses (one per Pod)
Get the NodePort either:
kubectl get service/rabbit --namespace=${NAMESPACE} --output=json
kubectl describe service/rabbit --namespace=${NAMESPACE}
You will need to use the NodePort to access both the MQTT endpoint and the UI.
statefulsets and deployments are different kubernetes resources. You have created statefulsets. That's why you don't see deployments. If you do
kubectl get statefulset you should see it and also both statefulset and deployment creates pod finally so you should be able to see rabbitmq pods if you do kubectl get pods
Since you have created a Nodeport service. You should be able to access it via http://nodeip:nodeport where nodeip is ip of any worker node in your kubernetes cluster.
You can get to know what is the Nodeport(a number between 30000-32767) by
kubectl describe services rabbit
Here is the doc on accessing a Nodeport service from outside the cluster.
In Dockerfile I have mentioned volume like:
COPY src/ /var/www/html/ but somehow my code changes don't appear like it used to only with Docker. Unless I remove Pods, it does not appear. How to sync it?
I am using minikube.
webserver.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: webserver
labels:
app: apache
spec:
replicas: 3
selector:
matchLabels:
app: apache
template:
metadata:
labels:
app: apache
spec:
containers:
- name: php-apache
image: learningk8s_website
imagePullPolicy: Never
ports:
- containerPort: 80
When your container spec says:
image: learningk8s_website
imagePullPolicy: Never
The second time you kubectl apply it, Kubernetes determines that it's exactly the same as the Deployment spec you already have and does nothing. Even if it did generate new Pods, the server is highly likely to notice that it already has an image learningk8s_website:latest and won't pull a new one; indeed, you're explicitly telling Kubernetes not to.
The usual practice here is to include some unique identifier in the image name, such as a date stamp or commit hash.
IMAGE=$REGISTRY/name/learningk8s_website:$(git rev-parse --short HEAD)
docker build -t "$IMAGE" .
docker push "$IMAGE"
You then need to make the corresponding change in the Deployment spec and kubectl apply it. This will cause Kubernetes to notice that there is some change in the pod spec, create new pods with the new image, and destroy the old pods (in that order). You may find a templating engine like Helm to be useful to make it easier to inject this value into the YAML.
Is possible to set a local image in a kubernetes pod yml file?
This is my pod yml file, and the question is if I can use a local image to containers tag (in local, I have all files to my api project, dockerfile, etc).
apiVersion: apps/v1
kind: ReplicaSet
metadata:
name: api-service
spec:
selector:
matchLabels:
api-name: api-service
replicas: 2
template:
metadata:
labels:
api-name: api-service
spec:
containers:
- name: api-service
image: #HERE
By local you mean it doesn't pull from dockerhub or any of the public registry. Yes it's possible if you run a single node kubernetes. You will utlize the docker cache where your kubernetes/kubelet is running.
First thing is, you need to set your imagePullPolicy: IfNotPresent. Then, when you build your image, you need to point to the docker instance your kubernetes is using.
I do this mostly with minikube, so the dev iteration is faster without pushing to my registry.
I've followed a few guides, and I've got CI set up with Google Container Engine and Google Container Registry. The problem is my updates aren't being applied to the deployment.
So this is my deployment.yml which contains a Kubernetes Service and Deployment:
apiVersion: v1
kind: Service
metadata:
name: my_app
labels:
app: my_app
spec:
type: LoadBalancer
ports:
- port: 80
targetPort: 3000
selector:
app: my_app
---
apiVersion: extensions/v1beta1
kind: Deployment
metadata:
name: my_app
spec:
replicas: 1
template:
metadata:
labels:
app: my_app
spec:
containers:
- name: node
image: gcr.io/me/my_app:latest
ports:
- containerPort: 3000
resources:
requests:
memory: 100
- name: phantom
image: docker.io/wernight/phantomjs:2.1.1
command: ["phantomjs", "--webdriver=8910", "--web-security=no", "--load-images=false", "--local-to-remote-url-access=yes"]
ports:
- containerPort: 8910
resources:
requests:
memory: 1000
As part of my CI process I run a script which updates the image in google cloud registry, then runs kubectl apply -f /deploy/deployment.yml. Both tasks succeed, and I'm notified the Deployment and Service has been updated:
2016-09-28T14:37:26.375Zgoogleclouddeploymentservice "my_app" configured
2016-09-28T14:37:27.370Zgoogleclouddeploymentdeployment "my_app" configured
Since I've included the :latest tag on my image, I thought the image would be downloaded each time the deployment is updated. Acccording to the docs a RollingUpdate should also be the default strategy.
However, when I run my CI script which updates the deployment - the updated image isn't downloaded and the changes aren't applied. What am I missing? I'm assuming that since nothing is changing in deployment.yml, no update is being applied. How do I get Kubernetes to download my updated image and use a RollingUpdate to deploy it?
You can force an update of a deployment by changing any field, such as a label. So in my case, I just added this at the end of my CI script:
kubectl patch deployment fb-video-extraction -p \
"{\"spec\":{\"template\":{\"metadata\":{\"labels\":{\"date\":\"`date +'%s'`\"}}}}}"
We have recently published a technical overview of how the approach that we call GitOps approach can be implemented in GKE.
All you need to do is configure GCR builder to pick-up code changes from Github and run builds, you then install Weave Cloud agent in your cluster and connect to a repo where YAML files are stored, and the agent will take care of updating the repo with new images and applying the changes to the cluster.
For a more high-level overview, see also:
The GitOps Pipeline
Deploy Applications & Manage Releases
Disclaimer: I am a Kubernetes contributor and Weaveworks employee. We build open-source and commercial tools that help people to get to production with Kubernetes sooner.
I have been using kubernetes for a while now.
$ kubectl version
Client Version: version.Info{Major:"1", Minor:"3", GitVersion:"v1.3.0+2831379", GitCommit:"283137936a
498aed572ee22af6774b6fb6e9fd94", GitTreeState:"not a git tree", BuildDate:"2016-07-05T15:40:25Z", GoV
ersion:"go1.6.2", Compiler:"gc", Platform:"darwin/amd64"}
Server Version: version.Info{Major:"1", Minor:"2", GitVersion:"v1.2.0", GitCommit:"5cb86ee022267586db
386f62781338b0483733b3", GitTreeState:"clean", BuildDate:"", GoVersion:"", Compiler:"", Platform:""}
I usually set an Ingress, Service and Replication Controller for each project.
apiVersion: extensions/v1beta1
kind: Ingress
metadata:
namespace: portifolio
name: portifolio-ingress
spec:
rules:
- host: www.cescoferraro.xyz
http:
paths:
- path: /
backend:
serviceName: portifolio
servicePort: 80
---
apiVersion: v1
kind: Service
metadata:
namespace: portifolio
name: portifolio
labels:
name: portifolio
spec:
selector:
name: portifolio
ports:
- name: web
port: 80
targetPort: 80
protocol: TCP
---
apiVersion: v1
kind: ReplicationController
metadata:
namespace: portifolio
name: portifolio
labels:
name: portifolio
spec:
replicas: 1
selector:
name: portifolio
template:
metadata:
namespace: portifolio
labels:
name: portifolio
spec:
containers:
- image: cescoferraro/portifolio:latest
imagePullPolicy: Always
name: portifolio
env:
- name: KUBERNETES
value: "true"
- name: BRANCH
value: "production"
My "problem" is that for deploying my app I usually do:
kubectl -f delete kubernetes.yaml
kubectl -f create kubernetes.yaml
I wish I could use a single command to deploy, whenever my app is up or down. Rolling updates do not work when I use the same image,(I think its a bug on my kubernetes server version). But it also do not work when the app has never been deployed at all.
I have read about Deployments, I wonder how it would help me?
Goals
1. Deploy if app is brand new
2. Replace existing pods with new ones using a new image from docker registry.
I don't think keeping all resources inside one single manifest helps you with what you want to achieve, since your Service, Ingress and ReplicationController are not likely to change simultaneously.
If all you want to do is roll out new pods, I would recommend you to replace your ReplicationController with a Deployment. Manifests have almost the exact same syntax so it's easy to migrate from standard RCs, and you could perform a server-side rolling update with a single kubectl replace -f manifest.yml.
Please note that even with a Deployment resource you can't trigger a redeployment if nothing changed in your manifest. kubectl replace would just do nothing. Therefore you could for example increment or change a tag inside your manifest in order to force the deployment, if needed (eg. revision: 003).
As already written in the previous answer, it is recommended to use a Deployment instead of a ReplicationController for this.
Using imagePullPolicy: Always will only ensure that Kubernetes does a docker pull before starting new PODs. It does not force recreation of PODs when nothing in the Deployment resource changes.
I would suggest to add 2 things to your solution:
Add a label to the Deployment with the value CURRENT_DATE as a placeholder value
Add a simple shell script to your project which replaces the placeholder with the current date+time and then uses kubectl to apply the resources.
Example Bash script
#!/usr/bin/env bash
sed "s/CURRENT_DATE/$(date)/" kubernetes.yaml | kubectl apply -f -
Then use this script for redeployment instead of calling kubectl by yourself.
This is only meant as a very simple example. When it comes to creating/applying/patching resources in Kubernetes, things tend to get more and more complicated by time. If this happens, consider using some more advanced templating solutions, e.g. by using Python and Jinja2.
You could use a deployment for this. Create it the first time, and after that you only need to do kubectl set image deploy/my-app app=user/image:tag --record and you're good to go.
Doing that, you can also do cool things like kubectl rollout undo deploy/my-app or get history and status.
You might consider using Argo.
Argo is an open-source workflow engine for Kubernetes. It allows to define complex microservices-based application deployment using YAML in source repo and automatically re-deploy app on YAML change (e.g. on every commit to production branch) .