Would there be any reason if I'm trying to run a pod on a k8s cluster which stays in the 'Completed' status forever but is never in ready status 'Ready 0/1' ... although there are core-dns , kube-proxy pods,etc running successfully scheduled under each node in the nodepool assigned to a k8s cluster ... all the worker nodes seems to be in healthy state
This sounds like the pod lifecycle has ended and most of the reasons is probably because your pod have finished the task it is meant for.
Something like next example will not do anything, it will be created successfully then started, will pull the image and then it will be marked as completed.
apiVersion: v1
kind: Pod
metadata:
name: my-pod
spec:
containers:
- name: test-pod
image: busybox
resources:Here is how it will look:
NAMESPACE NAME READY STATUS RESTARTS AGE
default my-pod 0/1 Completed 1 3s
default testapp-1 0/1 Pending 0 92m
default web-0 1/1 Running 0 3h2m
kube-system event-exporter-v0.2.5-7df89f4b8f-mc7kt 2/2 Running 0 7d19h
kube-system fluentd-gcp-scaler-54ccb89d5-9pp8z 1/1 Running 0 7d19h
kube-system fluentd-gcp-v3.1.1-gdr9l 2/2 Running 0 123m
kube-system fluentd-gcp-v3.1.1-pt8zp 2/2 Running 0 3h2m
kube-system fluentd-gcp-v3.1.1-xwhkn 2/2 Running 5 172m
kube-system fluentd-gcp-v3.1.1-zh29w 2/2 Running 0 7d19hFor this case, I will recommend you to check your yaml, check what kind of pod you are running? and what is it meant for?
Even, and with test purposes, you can add a command argument to keep it running.
command: ["/bin/sh","-c"]
args: ["command one; command two && command three"]something like:
args: ["-c", "while true; do echo hello; sleep 10;done"]The yaml with commands added would look like this:
apiVersion: v1
kind: Pod
metadata:
name: my-pod
spec:
containers:
- name: test-pod
image: busybox
resources:
command: ["/bin/sh","-c"]
args: ["-c", "while true; do echo hello; sleep 10;done"]Here is how it will look:
NAMESPACE NAME READY STATUS RESTARTS AGE
default my-pod 1/1 Running 0 8m56s
default testapp-1 0/1 Pending 0 90m
default web-0 1/1 Running 0 3h
kube-system event-exporter-v0.2.5-7df89f4b8f-mc7kt 2/2 Running 0 7d19h
kube-system fluentd-gcp-scaler-54ccb89d5-9pp8z 1/1 Running 0 7d19h
kube-system fluentd-gcp-v3.1.1-gdr9l 2/2 Running 0 122mAnother thing that will help is a kubectl describe pod $POD_NAME, to analyze this further.
This issue got nothing to do with k8s cluster or setup. It's mostly with your Dockerfile.
For the pod to be in running state, you need to have the container as an executable, for that usually an Entrypoint that specifies the instructions would solve the problem you are facing.
A better approach to resolve this or debug this particular issue would be, by running the respective docker container on a localmachine before deploying that onto k8s cluster.
The message Completed, most likely implies that the k8s started your container, then the container subsequently exited. In Dockerfile, the ENTRYPOINT defines the process with pid 1 which the docker container should hold, otherwise it terminates, so whatever the pod is trying to do, you need to check whether the ENTRYPOINT command is looping, and doesn't die.
For side notes, Kubernetes will try to restart the pod, once it terminates but after few tries the pod status may turn into CrashLoopBackOff and you will see message something similar to Back-off restarting failed container.