Now that we've moved away from Circle, here's how we develop our pipeline:

cd concourse
vagrant up
cd ..

grab fly from http://192.168.100.4:8080 (concourse may take a few minutes to come up locally) and put the binary in /usr/local/bin or somewhere else on your PATH

Now we can finish setting up concourse:

fly -t lite login -c http://192.168.100.4:8080
fly -t lite set-team -n main --basic-auth-username=foo --basic-auth-password=bar

Now we can deploy a pipeline! Isn't this what we've always wanted?

echo -e "concourse_url: 'http://192.168.100.4:8080'\nconcourse_user: foo\nconcourse_pw: bar" > local-concourse-config.yml
fly -t lite set-pipeline -p pgadmin -c pipelines/pgadmin4.yml -l local-concourse-config.yml
fly -t lite unpause-pipeline -p pgadmin

The first time we run we also need to create a no-op pipeline for the concourse-pipeline resource to be happy:

touch scratch.yml
fly -t lite set-pipeline -p pgadmin-feature-branches -c scratch.yml --non-interactive
fly -t lite unpause-pipeline -p pgadmin-feature-branches

For this open source project, the pipelines should be publicly accessible.

fly -t lite expose-pipeline -p pgadmin
fly -t lite expose-pipeline -p pgadmin-feature-branches

Now go back to your concourse and kick off the manually-generate-feature-branch-pipeline job

If that succeeds, the pgadmin-feature-branches pipeline will now contain several jobs (as specified in pipelines/feature-branch-pipeline-jobs.yml.erb) for each branch.

You can find a job named [your feature branch of interest]-branch and trigger that. The corresponding jobs will then run automatically (triggered by new commits pushed to that branch).

$ docker build -t pivotaldata/pgadmin:postgres-10 .

$ docker tag postgres-10 pivotaldata/pgadmin:postgres-10

$ docker push pivotaldata/pgadmin:postgres-10

In a single command:

export TAG=postgres-10 && docker build -t pivotaldata/pgadmin:$TAG docker/$TAG && docker push pivotaldata/pgadmin:$TAG

$ ./fly -t lite intercept -j pgadmin/Python-3.6-Postgres bash

The following sections may still come in handy as part of concourse setup or for pushing manually?

Push a branch from the pgadmin4 repo, which will trigger a git hook that updates this repository and it's submodules.

Then login to CF using the following command:

$ cf login -a https://api.run.pivotal.io
> USERNAME
> PASSWORD

Run the script that will push:

$ ./push_cf.sh

If the following message is displayed:

Not deploying because couldn't find the branch name in the git message

Check the last git message, it should look like this:

[branch-name-id] Run tests for.....

If it doesn't, the commit was not generated by the Git Hook in the pgadmin4 repo and will not be pushed to CF