K
Q

Need a CURL binary availble inside kubernetes pod

11/13/2018

I would like to sh myself inside a kubernetes pod and execute a CURL command. Unfortunatly I can't find anywhere a working image with curl availble (and compatible with kubernetes)...

  1. I tried some docker images with Alpine and CURL but each time it ended with crashLoopBackOff. I guess it means the container exited because the docker image exits after executing itself...
  2. I also tried using the image of alpine and ubuntu alone but each time it also ended with crashloopBackOff.
  3. I manage to exec in a few images but it never had CURL installed and neither APT-GET or APK were working.

To exec into a container I'm doing a simple kubectl exec -it POD_ID /bin/bash

Does someone knows of a minimal docker image that contains a CURL binary and wont crash in kubernetes ?

PS: This is for testing purpose so it does not need to be rock solid or anything

Thx

UPDATE 1 This is the yaml I use to deploy all potential image :

apiVersion: apps/v1beta1
kind: Deployment
metadata:
  name: blue
  namespace: default
spec:
  replicas: 1
  template:
    metadata:
      labels:
        name: blue
    spec:
      containers:
      - name: blue-website
        image: SOME_IMAGE:latest
        resources:
          requests:
            cpu: 0.1
            memory: 200

I don't think that its broken because it works on certain image.

-- Doctor
alpine
curl
docker
kubernetes

Similar Questions

Error from server (BadRequest): container "mariadb" in pod is waiting to start: CreateContainerConfigError
What are the equivelant kubectl commands to this yaml?
How Can You Perform Variable Substitution using Kubectl?
GCP Bucket gcfuse and Kubernetes Engine rights
Can't load Emacs kubel.el on Emacs 27.2 on Macbook Pro

4 Answers

11/13/2018

You can skip the manifest and use kubectl run to spin up one of these pods on demand. i.e.

kubectl run curl -it --rm --image=tutum/curl -- sh

This would create a deployment named curl from the tutum/curl image and give you an interactive (-it) shell inside it. When you exit, the deployment will be deleted (--rm).

-- switchboard.op
Source: StackOverflow
11/13/2018

You getting CrashLoopBackOff because container completes after starting as it does not have any task to process. Easy workaround is to run a command in the container to keep it running indefinitely. So that you can exec into the container and run curl.

Here, is modified yaml to do this:

apiVersion: apps/v1beta1
kind: Deployment
metadata:
  name: blue
  namespace: default
spec:
  replicas: 1
  template:
    metadata:
      labels:
        name: blue
    spec:
      containers:
      - name: blue-website
        image: scrapinghub/httpbin:latest
        command:
        - sleep
        - "3600"
        resources:
          requests:
            cpu: 0.1
            memory: 200
-- Emruz Hossain
Source: StackOverflow
11/13/2018

You can use this image nightfury1204/alpine-curl

I created above image for my own testing purpose.

apiVersion: apps/v1
kind: StatefulSet
metadata:
  name: curl
  labels:
    name: curl
spec:
  serviceName: "curl"
  selector:
    matchLabels:
      app: curl
  replicas: 1
  template:
    metadata:
      labels:
        app: curl
    spec:
      containers:
      - name: curl
        image: nightfury1204/alpine-curl
        command:
          - "sh"
          - "-c"
          - >
            while true; do
              sleep 3600;
            done

To exec into the pod use this kubectl exec -it curl-0 sh

-- nightfury1204
Source: StackOverflow
11/14/2018

Use byrnedo/alpine-curl image from https://hub.docker.com/r/byrnedo/alpine-curl/. Also it's not neccessary to have latest tag in the deployment. It works without it, just

containers:
  - name: blue-website
    image: byrnedo/alpine-curl
-- Anna Slastnikova
Source: StackOverflow