Incorrect pod state shown

[hookenz]

Before you start please confirm the following.

• Yes, I've searched similar issues on GitHub.
• Yes, I've checked whether this issue is covered in the Portainer documentation or knowledge base.

Problem Description

When I run rollout restart on a deployment, to ensure the correct uptime guarantee one container will be created and all traffic routed to it, while the container it replaced is terminated.

kubectl is showing the correct status.
However, Portainer is showing both containers in the running state

Expected Behavior

The terminating container should show "terminating".

Actual Behavior

It doesn't, it shows "running"

Steps to Reproduce

A deployment that takes a while to get going and to terminate.
Image pull policy is set to "always" in my case and tag = "latest"
Replica Count is 1.

kubectl rollout restart deployment/your-deployment
kubectl get all

compare to portainer deployment app view showing running pods

Portainer logs or screenshots

I have a screencast of it but cannot seem to update it. Here is a screenshot of it. Notice the state of the pods.
I refreshed the portainer screen after kubectl was already showing "terminating". You can clearly see the pod state still shows "running"

Portainer version

2.31.3

Portainer Edition

Community Edition (CE)

Platform and Version

microk8s

OS and Architecture

Ubuntu 22.04

Browser

firefox

What command did you use to deploy Portainer?

standard

Additional Information

No response

[Nick-Portainer]

@hookenz Do you have a manifest we can use as an example to keep this as close as we can to your example? I am having issues with "A deployment that takes a while to get going and to terminate."

[hookenz]

@Nick-Portainer this does it.

apiVersion: apps/v1
kind: Deployment
metadata:
  name: debian-bash
spec:
  replicas: 1
  selector:
    matchLabels:
      app: debian-bash
  template:
    metadata:
      labels:
        app: debian-bash
    spec:
      containers:
        - name: debian
          image: debian:latest
          imagePullPolicy: Always
          command: ["bash", "-c", "while true; do sleep 3600; done"]

And here is the issue. It should be obvious when looking at this function in app/kubernetes/pod/converter.js
This is called when displaying this table.

portainer/app/kubernetes/pod/converter.js

Lines 29 to 42 in 3bf84e8

	function computeContainerStatus(statuses, name) {
	const status = _.find(statuses, { name: name });
	if (!status) {
	return 'Terminated';
	}
	const state = status.state;
	if (state.waiting) {
	return 'Waiting';
	}
	if (!state.running) {
	return 'Terminated';
	}
	return 'Running';
	}

Basically if its not terminated or waiting then it's going to return Running.

However it turns out that kubectl get pods which shows a status is a bit more complicated than just looking at the status field.

I think the logic from kubectl pod status is buried in here somewhere

https://github.com/kubernetes/kubectl/blob/master/pkg/describe/describe.go

Maybe something like this will work? (note: Generated by AI so not validated at all!)

function computeContainerStatus(statuses, name) {
  const status = statuses.find(s => s.name === name);
  if (!status) {
    return 'Unknown';  // Better fallback
  }
  const state = status.state;

  if (state.waiting) {
    // Return the actual waiting reason, or default to 'Waiting'
    return state.waiting.reason || 'Waiting';
  }
  if (state.terminated) {
    // Check exit code for success/failure
    if (state.terminated.exitCode === 0) {
      return 'Terminated';
    }
    // Non-zero exit - could be 'Error' or 'CrashLoopBackOff'
    return state.terminated.reason || 'Error';
  }
  if (state.running) {
    // Optionally check if container is ready or not
    return status.ready ? 'Running' : 'Starting';
  }

  return 'Unknown';
}