Pod failure with custom docker image

[ruben-bc]

Hi all,

I'm doing a simple web app that consists in a visitor counter every time the page is reloaded. To do it I have a script in JavaScript and I need to connect it to a HTML file which acts as the web by itself, and the base image I want is nginx. As I understood, the index.html of it cannot be edited, so it's necessary to build a custom docker image (docker build -t custom-nginx .) with new files and a Dockerfile (starting from the original nginx image).

My Dockerfile:

FROM nginx:latest
COPY ./index.html /usr/nginx/share/html/index.html
COPY ./index.js /usr/nginx/share/html/index.js
COPY ./styles.css /usr/nginx/share/html/styles.css

When I run this separately (docker run...) it works fine, but when I configure it in the platform, almost with the same configuration as for nginx (only changing the image container name), it doesn't work, pods fail.

kubectl get pods -n nginx-sandbox:

NAME                                   READY   STATUS             RESTARTS   AGE
meep-ams-667d8fd654-j49vh              1/1     Running            0          112m
meep-app-enablement-7bc99dc6d4-nkzn6   1/1     Running            0          111m
meep-gis-engine-7fdc57f844-nx7cr       1/1     Running            0          112m
meep-loc-serv-59dd45c974-4xhg8         1/1     Running            0          112m
meep-metrics-engine-679847c579-pvbh2   1/1     Running            0          112m
meep-mg-manager-76b77554d6-n2dwz       1/1     Running            0          111m
meep-rnis-765f646c59-4gq4p             1/1     Running            0          111m
meep-sandbox-ctrl-7bb48c5bc5-hwcjp     1/1     Running            0          112m
meep-tc-engine-99c8cdc7c-dfsnx         1/1     Running            0          112m
meep-wais-5dbd6f9986-bxh29             1/1     Running            0          111m
nginx-cloud-f858cc548-wmgxr            1/2     ImagePullBackOff   0          25m
nginx-edge-6dbb94d746-952v4            1/2     ImagePullBackOff   0          25m
nginx-fog-744f85765-zvkt9              1/2     ImagePullBackOff   0          25m
ue1-app-59c7d4df95-7sbw9               1/1     Running            0          25m

I checked the events in one these pods (kubectl describe pod nginx-edge-6dbb94d746-952v4) and I saw that error:

Events:
  Type     Reason     Age                  From                 Message
  ----     ------     ----                 ----                 -------
  Normal   Scheduled  2m21s                                     Successfully assigned nginx-sandbox/nginx-edge-6dbb94d746-952v4 to advantedge
  Normal   Pulled     2m15s                kubelet, advantedge  Container image "busybox:1.28" already present on machine
  Normal   Created    2m14s                kubelet, advantedge  Created container init-kube-dns
  Normal   Started    2m12s                kubelet, advantedge  Started container init-kube-dns
  Normal   Pulling    2m9s                 kubelet, advantedge  Pulling image "meep-docker-registry:30001/meep-tc-sidecar:latest"
  Normal   Pulled     2m9s                 kubelet, advantedge  Successfully pulled image "meep-docker-registry:30001/meep-tc-sidecar:latest" in 672.276856ms
  Normal   Created    2m8s                 kubelet, advantedge  Created container meep-tc-sidecar
  Normal   Started    2m7s                 kubelet, advantedge  Started container meep-tc-sidecar
  Normal   Pulling    29s (x3 over 2m12s)  kubelet, advantedge  Pulling image "custom-nginx"
  Warning  Failed     24s (x3 over 2m9s)   kubelet, advantedge  Error: ErrImagePull
  Warning  Failed     24s (x3 over 2m9s)   kubelet, advantedge  Failed to pull image "custom-nginx": rpc error: code = Unknown desc = Error response from daemon: pull access denied for custom-nginx, repository does not exist or may require 'docker login': denied: requested access to the resource is denied
  Normal   BackOff    11s (x4 over 2m6s)   kubelet, advantedge  Back-off pulling image "custom-nginx"
  Warning  Failed     11s (x4 over 2m6s)   kubelet, advantedge  Error: ImagePullBackOff

Also I noticed that there's not Container ID and Image ID tagged inside pod description, but with docker images can see that it has an ID for the image.

I think that is a configuration problem between custom image and k8s, because the base image is nginx and with the original it works, the pods run and show the webpage entering to the proper URL. I've checked the external SVC server of demo1 in case I could understand what could be happening. Still, I'm not able to see specifically what the fault is.

Thank you in advance for the help.

[roymx]

Hello @rubenbc28 ,

This seems like a docker registry problem.
K8s seems unable to access your docker image called custom-nginx in the specified docker registry.
1- What docker registry did you specify in your scenario?
2- Make sure the docker-registry is accessible from K8s

More details...
After dockerizing your image, it will be present in the docker registry of the host where you dockerized.
If this docker registry is not accessible to K8s then it will not be able to pull the container.

• nginx official image is probably accessible in a cloud registry, most certainly accessible from K8s
• AdvantEDGE containers are pushed from the host computer docker registry to the meep-docker-registry container available within the platform - so these are certainly accessible from K8s

To see how to push images from the local host to the meep-docker-registry, try this meepctl dockerize all -v | grep push

Command to push your image should look like this docker push meep-docker-registry:30001/custom-nginx
Then in your scenario you would need to indicate the container name as meep-docker-registry:30001/custom-nginx

Let me know if it worked

[ruben-bc]

Hi Mike,

Due to dockerize only works with the valid targets referenced in https://github.com/InterDigitalInc/AdvantEDGE/blob/master/docs/meepctl/meepctl_dockerize.md, I did directly the docker push command and worked. Now I can use the custom image like any original.

Thanks!

[Acacio98]

Hi Rubenctl28, I am also deploying an application of my own for use with the AdvantEdge platform. I am starting with a simple web application. It has three files (.css .html .js), then I have my Dockerfile, running it separately to the platform works for me, but I would like to configure it on the platform. What did you change or add in the platform files for the correct functioning of your application.

Thanks in advance, I look forward to your reply.

[ruben-bc]

Hi @Acacio98,

As described above, it's necessary to push your custom Docker image in the private registry of the platform, instead of the host local registry. By tagging your image (docker tag <image_name> meep-docker-registry:30001/<image_name>) and then pushing it (docker push meep-docker-registry:30001/<image_name>), you can then reference that container image name in your scenario configuration.

Hope it helps.

[Acacio98]

Hi @rubenbc28, thanks for the answer and the quick response, it was very helpful. But I have a couple more short questions. in the Dockerfile to create the image, do you specify a port with the "EXPOSE" command.

Did you get to work the small application of the hit counter, if it was not too much to ask, if you could pass me the files to see if I solve any doubt.

Thank you very much again.

[ruben-bc]

Hi @Acacio98,

If you want to expose one or more ports, you have to configure it in a terminal application as exemplified in Demo 1 for UE2, and previously make sure that the port you are going to use is specified in the network elements that you need. Docker images are just a means for AdvantEDGE to create the container and run the contained applications, so generally what you configure if you want a custom image should be relative to how you want your application to be, because the behavior that it will have at the network is modeled on a scenario in the platform interface.

This should solve your question.