Using variable in environment name

How to Use GitLab
,
1

I have the following configuration:

    ...
    variables:
      NAMESPACE: '$CI_COMMIT_REF_SLUG'
    deploy:
      script: ...
      environment:
        name: ci/$NAMESPACE
   ...

According to documentation I do not see a limitation:
https://docs.gitlab.com/ee/ci/variables/where_variables_can_be_used.html
https://docs.gitlab.com/ee/ci/environments/#allowed-variables

But when running it shows ci/$CI_COMMIT_REF_SLUG in the jobs and environment view. While I expect e.g. ‘ci/master’. The script itself seems to work fine, but I can not make distinction between different runs on different branches. Anyone an idea why this occurs?

version: GitLab Community Edition [13.8.2]

1 Like
2

I think that’s an error. Here’s a quick and dirty testing repository that demonstrates the issue:
https://gitlab.com/gitlab-test-submodule-mergetrain/test-ci-env-var/-/environments
Using variables defined even on top of .gitlab-ci.yml do not work.

However the sample from in the documentation does work:

    environment:
        name: env4/$CI_COMMIT_REF_NAME
3

Thanks for confirming that it is an error. How to proceed with this? Would like to have it fixed or have a workaround :slight_smile:

4

I wonder whether it’s because you are using $CI_COMMIT_REF_NAME rather than $CI_COMMIT_REF_SLUG which escapes the text in the commit name, and is safe to use in URLs?

Regards,

Sarah

5

If I do it with an indirect reference through variables as in my example both does not work.

6

Perhaps I’ve misunderstood your question? This is how I do a similar thing:

deploy:
    stage: deploy
   ...
    script:
        - ...
    rules:
        - if: '$CI_COMMIT_BRANCH == "master"'
          when: never
        - when: always
    artifacts:
        paths:
            - public
    environment:
        name: ci/$CI_COMMIT_BRANCH
        url: ...
        auto_stop_in: ...

Does that help?

Sarah

7

Nope :slight_smile: I want to use a variable in environment.name. Reason is that I want to trigger the pipeline and give as variable input the environment I want to create and the default must be the branch name itself.

8

I think the config above does set the name with a variable?

9

Yes, but CI_COMMIT_BRANCH is a predefined variable which I can not provide during start of the pipeline. The example I provided is exactly what I need.

10

Sorry, @dkwakkel I’m obviously very confused. What’s the reason for giving CI_COMMIT_REF_SLUG an alias? Normally we would just use those variables as they are in the CI configs?

11

See my comment above:

12

Right, apologies. I suspect your problem stems from this issue which is unfortunately still open :frowning:

Sarah

13

Thanks Sarah!

14

This issue seems to still be a problem. The predefined variables don’t seems to work in environment names even though the docs say they do.

image

  environment:
    name: $ENVIRONMENT/$APP_NAME
  variables:
    ENVIRONMENT: test
    APP_NAME: build-$CI_BUILD_REF_SLUG

Update:

This is a variable in variable issue since the following does work correctly:

  environment:
    name: $ENVIRONMENT/$APP_NAME-$CI_BUILD_REF_SLUG
  variables:
    ENVIRONMENT: test
    APP_NAME: build
1 Like