Building and Pushing Docker Images

[Samuel Knowlton]

(question moved from the Bitbucket section)

We have a simple app we're trying to build in much the same vein as this article:

https://confluence.atlassian.com/bitbucket/build-and-push-a-docker-image-to-a-container-registry-884351903.html

We have everything the app needs except the repo on a private image in GCR.  We've set up a pipelines yml as close to the example as we can get, but when it gets to the docker build portion, it throws this error:

unable to prepare context: unable to evaluate symlinks in Dockerfile path: lstat /opt/atlassian/pipelines/agent/build/Dockerfile: no such file or directory

What we want to happen here is for Pipelines to take our base image (image A), clone the repo into a folder on that image, and then push the resulting image back up as a production-ready image (image B, which is just image A with our repo in it)

I'd like to confirm my understanding of how this would occur:

- the Pipelines Docker build environment is pulling our image and (as the build setup process) creating a container from that image and cloning the repo into that container in  $BITBUCKET_CLONE_DIR

- the build environment (not our container) is then building (or committing?) the container that now has our repo and pushing it up 

options: docker: true
image:  name: us.gcr.io/our-project/ourImage:buildTag
username: _json_key
password: '$GCLOUD_API_KEY'
pipelines:
  branches:
    Production: 
     - step:
       name: Build Production Image
       deployment: production
       caches: - docker
       script:
         - docker build -t us.gcr.io/$GCLOUD_PROJECT/$IMAGE_NAME:$IMAGE_TAG . 
         - docker push us.gcr.io/$GCLOUD_PROJECT/$IMAGE_NAME:$IMAGE_TAG

I think what I'm confused about is the Dockerfile: am I meant to supply a Dockerfile in that path on my image? Or is docker build not even what I'm after here since I just want to commit the environment as I have it?

Thanks for your help!

[Philip Hodder]

Hi Samuel,

What is the location of your Dockerfile? The error is indicating that Docker can't find a Dockerfile in the current directory.

If you need to specific the path of the Dockerfile, you can use the `-f` flag

docker build -f <path_to_docker_file> -t us.gcr.io/$GCLOUD_PROJECT/$IMAGE_NAME:$IMAGE_TAG . 

Thanks,

Phil

[Samuel Knowlton]

Hi Phil,

I know that I can supply a Dockerfile, but the example (in the link in my post) didn't reference doing that -- I figured out from the error that I could just add the folder /opt/atlassian/pipelines/agent/build/ to my build image and put a Dockerfile in it, but the help article didn't say anything about that.

So while on the one hand it's obvious that you need a Dockerfile for docker build to do anything -- the Atlassian page says "just do this and Pipelines will build your app" while running their code exactly will produce the error. I wasn't sure when I wrote this post what all was going on in Pipelines (since it has its own Docker environment) and at first it seemed like they were injecting a Dockerfile based on Pipelines environment variables. That made less sense as I got into it.

Thanks for your assistance!

Sam

[Murali Krishna Dupati]

Any one got solution?

Do we need add our own Docker file ?

Thanks

Murali

[Oleg Sigida]

sure you need to create and commit Dockerfile, as only you knows how image should be build.

Bitbucket only a scheduler who executes predefined script, nothing more.

[pushpender singh]

Any one got solution?

Do we need add our own Docker file ?

[Philip Hodder]

Yes, you need to have your own Dockerfile. You just need to run docker commands in the same way you would locally.

[Cristhian Benitez]

I do have a Dockerfile and I have the same issue:

unable to prepare context: unable to evaluate symlinks in Dockerfile path: lstat /opt/atlassian/pipelines/agent/build/Dockerfile

[murchd]

You either have no Dockerfile in the current directory or your Dockerfile is called something else. You can specify the docker file with the -f option.

$ docker build -f Dockerfile.debug .

For more info

https://docs.docker.com/engine/reference/commandline/build/