Debug School: Nagendra

Services Resources in K8s

Nagendra — Thu, 19 Oct 2023 04:31:32 +0000

Basically Services means the Network load balancer.

There will be a problem like if some one has to call the rest call where inside the POD its running in that case he needs the ipaddress and POrt.

But these ip address keeps on changing when the pod fone k8s will create one more pod with different ip.

In this case we have the use the Services Basically this will works based on the labels

" POd label should match with the services selector label " then those Services provides one of the ip with that ip we can access that "

like --type=clusterIp : kubectl expose rs nagendrareplica --port=1234 --target-port=80

There are three types:

Clusterip : We can use this cluster ip to access tha application inside the cluster with the port not outside of it
http://10.100.6.37:80
--type=nodeport : This one will do the cluster ip step and also opens the node port which can be accessible by outside
root@ip-172-31-46-170:/home/ubuntu/nagendra# kubectl expose rs nagendrareplica --port=1234 --target-port=80 --type=NodePort -n=nagendra
service/nagendrareplica exposed

From Clusterip with port we can access inside the cluster
From nodeip where services is running with the port given by node port we can access outside whcih will redirect that traffic into the correspoding application

type=--loadbalancer : This will create the loadbalancer ip and the node port. Here k8s not able to create the loadbalancer thats why its coming as pending --type=LoadBalancer : ubectl expose rs nagendrareplica --port=1234 --target-port=80 --type=LoadBalancer -n=nagendra servicel1 LoadBalancer 10.111.196.233 1234:30649/TCP 5s

Workflow like : f user do the curl of www.app.xxx. ( which was added in DNS mapping there we have to mention the loadbalancerip and node port)

Then the request goes to the loadbalancer ip to the one of the node and that corresponding nodeport and then it redirected that info to the 80 port and gives the result

Note : There is a problem like if there are too many services resources like if its 100 then it has to create 100 loadbalancer which is costlier.

Exapmples :

root@ip-172-31-46-170:/home/ubuntu/nagendra# kubectl get services -n=nagendra
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
nag-cs ClusterIP 10.100.6.37 5678/TCP 66m
nagendrareplica NodePort 10.100.64.125 1234:30710/TCP 25s
nagendrareplicav2 ClusterIP 10.111.113.155 80/TCP 56m

Deamon Sets, Jobs , Cron Jobs

Nagendra — Thu, 19 Oct 2023 04:16:48 +0000

DeamonSets :
This deamon sets creates only one POD in all the nodes not more than that Which was managed by the deamon controller (Here controller make sure all the nodes contains one pod of it)

We can create deaomnsets using the yaml file like below (kind: DaemonSet)

apiVersion: apps/v1
kind: DaemonSet
metadata:
name: logging
spec:
selector:
matchLabels:
app: httpd-logging
template:
metadata:
labels:
app: httpd-logging
spec:
containers:
- name: webserver
image: httpd
ports:
- containerPort: 80
In the same we can descibes , edit , delete the deamset resources
LIke we can list

kubectl get ds (ds is a shortname of deamon sets)

JOBS/ CronJobs :
JObs is a resources which used only to run do the task and done with that POD

CronJobs : This one scheduling that POD every 1 hour u execute the container like that.

kubectl get jobs
2030 kubectl get pods
2031 kubectl apply -f job.yaml
2032 kubectl get pods
2033 clear
2034 kubectl describe job pi
2035 clear
2036 vi cron.yaml
2037 kubectl apply -f cron.yaml
2038 kubectl get cronjobs
2039 kubectl get pods
2040 clear
2041 kubectl get cronjobs
2042 kubectl get pods
2043 kubectl get cronjobs
2044 kubectl describe cronjob hello

Config Map:

This config map which stores in the k8s cluster whcih is in the format of key-value pair (Which can be accessible by all in the cluster)

If we want this in pod please attach to the POD and mount to the container then we can use that key-value inside the container

Commands : ls
2055 vi reverseproxy.conf
2056 kubectl create configmap my-config --from-file=reverseproxy.conf
2057 kubectl get cm
2058 kubectl describe cm my-config
2059 clear
2060 ls
2061 vi cm.yaml
2062 kubectl apply -f cm.yaml
2063 kubectl get pods
2064 history
2065 kubectl exec helloworld-nginx ls /etc/nginx/conf.d
2066 kubectl exec helloworld-nginx more /etc/nginx/conf.d/myconfo.conf
2067 history

Working With Namespace,POD,ReplicationController and Deployment

Nagendra — Wed, 18 Oct 2023 04:31:19 +0000

NameSpace : (its kind of resources) (logical seperatin of cluster

There are situation like each k8s cluster used by multiple teams. If one team container consuming more memory than it affects the other team also and also every team member can access all the team pods which is not correct

On that time we will use the namespace one where each team can create there own process and allocate there resources (like it cant consume more than some amount of CPU and memory)

To create a NameSpace :
kubectl create ns ex kubectl create ns nagendra (which creates the namespace call nagendra)

To describe the namespace :
kubectl describe ns nagendra

To edit the namespace
kubectl edit ns nagendra

To delete the namespace
kubectl delete ns nagendra

TO list all the namespace
kubectl get ns

Note : For all the resources we use the CRUD operations are common

POD :

POD is a logical unit which cant be created. We can instantiate the POD (which create a own network space like port space)

Write pod.yaml file (Attach the content of pod.yaml once we got the access)

To create a POD :
kubectl create -f pod.yaml ( whcih creates the pod mention in the pod.yaml file here ex nagendrapod)

To describe the POD:
kubectl describe pod nagendrapod

To edit the pod
kubectl edit pod nagendrapod (Which edit directly the eecd)

To delete the pod
kubectl delete -f pod.yaml

TO list all the pod
kubectl get pod

kubectl get pods -o wide (Shows the ip address fll details of the pods)

TroubleShoot Of PODS ( Each pod contains multiple containers>

kubectl logs --- > to get the logs of the pod
kubectl attach --> Which will attach to the pod to the PID1 (like docker)
kubectl exec ls --> Execute the command internal to the POD (like docker)
kubectl exec -it /bin/bash --> Interactive mode the POD
kubectl port-forward --address 0.0.0.0 pod/rajesh 8888:80 (Explaination) --address (Any address) pod/nameofthepod nodeport:containerport (Explaination of the above command) Any address can communicate with the pod (0.0.0.0 means) if it outside guy send data to the 8888 it forwards to 80)

kubectl auth can-i ( give the persmission of user )

kubectl auth can-i create ns (If the outpput is true ) then user have the access to create NS

Replication controller : (Another resources Replication means 1 to many and controller means it watches and maintains the desired state mentioned in the .yaml)

apiVersion: v1
kind: ReplicationController
metadata:
name: nagendrarc
spec:
replicas: 5
template:
metadata:
labels:
app: nginx
spec:
containers:
- name: nginx
image: scmgalaxy/nginx-devopsschoolv1

kubectl scale --replicas=1 nagendrarc ( in above once we execute it creates 5 pods that image After that we execute this command then pod value come to 1)

kubectl create -f rc.yaml
kubectl get rc
kubectl edit rc nagendrarc
kubectl get rc nagendrarc -o yaml (Which gives the yaml file )
kubectl delete rc nagendrarc

https://stackoverflow.com/questions/43147941/allow-scheduling-of-pods-on-kubernetes-master

NOw replicationController is deprecated instead of that we use the resource ReplicaSet means in .yaml file we use the ReplicaSet

Same command as replication controller

Deployment (Another resource most of the time we use this one)

We can do the deployment without yaml file like below

kubectl create deployment --image=scmgalaxy/nginx-devopsschoolv1 --replicas=5

Which will create the 5pods where that image container is running

kubectl describe deploy my-dep (which will describes the deployment)

kubectl scale --replicas=2 deploy/my-dep ( which will bring back the replica 5 to 2)

kubectl rollout history dep/my-dep ( whcih will tell the version of the deployment)

Suppose there is a situation we have to upgrade to higher version then it will be easy using this

kubectl patch deployment my-dep --type='json' -p='[{"op": "replace", "path": "/spec/template/spec/containers/0/image", "value":"scmgalaxy/nginx-devopsschoolv2"}]' ( Patch apply the patch )
kubectl rollout status deploy/my-dep ( which will status of the rollout)

IF we want to go back to previous version of deployment (means basically previous image in PODS ) it will be easy
kubectl rollout undo deploy/my-dep --to-revision=2

What is POD?

Nagendra — Mon, 16 Oct 2023 11:48:07 +0000

POD cant create its only can instantiate by kubetl
EaCh Node can contain multiple pods
Each pod can contains multiple container (or) single containers
If container is not there then POD also not there viceversa
POD is the logical unit (not like docker and all)
kubernetes will only schedulde the pods not others
Ideal design is each pod contains single container not more than that. (We can add but its not ideal)
Internally POD means containers in pods communicate with through localhost
When 2 pods communicates it will go through the pod network

Kubernetes From Nagendra

Nagendra — Mon, 16 Oct 2023 07:07:12 +0000

What is Kubernetes? Kubernetes invented from google but now managing by the CNFC (opensource) which manages/orchestrator the dockers in a physical/ virtual machines
why do we need kuberenetes? Kubernetes solves the below problems a. It will manage both the work (dockers ) and workers (node) in efficient manner. If it manage by the ppl takes lot of time but it does in a seconds. b. Only dockers manages on one node if its multiple nodes then this k8s will take care
How Kubernetes Works? and architecture ? K8s has master and worker nodes All the k8s communication goes through the master nodes (from outside world to the minion (node)) while managing the server Each master nodes has 4 components
1. Apiserver -- > Worker nodes and the workstation (who given the instrucation) can talk only through the api server
2. CLuster store --> Which stores all the information (Which is SOT) related to Worker nodes (like how many pods and there state information everything)
3. Controller --> Which looks up the status of the nodes and the pods and sending that information to the apiserver and resposible to manintain the desired state
4. Schedular --> Which schedules the work to nodes has given by the apiserver (like which pod has to run in which node ) and all.