I am trying to execute a job on some pipeline variables. I have used 'rules' in my .gitlab-ci.yml file but getting the error "key may not be used with 'rules': only".
How can I do this?
build-dev:
stage: build
only:
- master
- branches
rules:
- if: '$CI_COMMIT_BRANCH=="my-featured-branch"'
when : never
The error you're receiving is literally what it says: you shouldn't use only with rules together in the same job.
Basically the reason is that this could lead into problems due to mixed behavior.
From the documentation:
rules replaces only/except and they can’t be used together in the same job. If you configure one job to use both keywords, the GitLab returns a key may not be used with rules error.
Related
I saw this code on one of the pipelines of my company.
rules:
- if: 'SOME-CONDITION'
when: manual
- when: never
variables:
...
According to gitlab ci documentation, the when:never field should be used with a condition, to basically tell the pipeline to not add the job if that condition is satisfied. I don't understand its use by itself in the end of the rules. What does it add and how the pipeline will behave without it ?
The last when: never is not needed.
Even without that line, the job will run only if SOME-CONDITION is satisfied.
How do I add a label to the GitLab pipelines when they run?
This would be extremely helpful when you run a few nightly (scheduled) pipelines for different configurations on the main branch. For example, we run a nightly main branch with several submodules, each set at a point in their development (a commit point SHA) and I want to label that 'MAIN'. We run a second pipeline that I want to label 'HEADs', which is a result of pulling all of the HEAD's of the submodule to see if changes will break the main trunk when they are merged in.
Currently it shows:
Last commit message.
Pipeline #
commit SHA
Branch name
'Scheduled'
That is helpful, but it is very difficult to tell them apart because only the pipeline # changes between the pipelines.
I have good news!!
Our friends at GitLab have been working on this feature. There is now a way to label your pipeline in release 15.5.1-ee.0!
It uses the workflow control with a new keyword name
workflow:
name: 'Pipeline for branch: $CI_COMMIT_BRANCH'
You can even use the workflow:rules pair to have different names for you pipeline:
variables:
PIPELINE_NAME: 'Default pipeline name'
workflow:
name: '$PIPELINE_NAME'
rules:
- if: '$CI_PIPELINE_SOURCE == "merge_request_event"'
variables:
PIPELINE_NAME: 'MR pipeline: $CI_COMMIT_BRANCH'
- if: '$CI_MERGE_REQUEST_LABELS =~ /pipeline:run-in-ruby3/'
variables:
PIPELINE_NAME: 'Ruby 3 pipeline'
Find the docs here: https://docs.gitlab.com/ee/ci/yaml/#workflow
This feature is disabled by default in 15.5 because it is so new.
You can enable the feature flag, which is named pipeline_name.
See this link to enable: https://docs.gitlab.com/ee/administration/feature_flags.html
(You need to use the Rails Console to enable it. Pretty easy.)
Note: Remember that the workflow keyword affects the entire pipeline instance.
This seems to be officially supported with GitLab 15.7 (December 2022)
Add custom names to pipelines with workflow:name:
For some projects, the same pipeline can be configured to run differently for different variables or conditions, creating very distinct outcomes for successful pipelines.
It can be hard for you to determine which version of that pipeline ran since there is no indication about the inputs used for that particular run.
While labels like scheduled and API help, it is sometimes still difficult to identify specific pipelines.
Now you can set a pipeline name using the keyword workflow:name to better identify the pipeline with string, a CI/CD variable, or a combination of both.
See Documentation and Issue.
Note:
If the name is an empty string, the pipeline is not assigned a name.
A name consisting of only CI/CD variables could evaluate to an empty string if all the variables are also empty.
I have a few use cases in my Gitlab setup I would like to be able to support:
If a certain label (let's call it “skip_build”) is set, the deployment steps should not be run when I merge an MR to a main branch. This would be useful when we have multiple MRs being merged right after another and only need the last one built.
If another label (we'll call it “skip_tests”) is set, I should be able to read it as an env var from within the script and alter the flow within the script accordingly (using normal bash syntax), e.g. to alter the package command parameters used a bit. This is useful for small changes where it might not make sense to run a lengthy test suite.
Is this possible with Gitlab, and if so, how?
I’ve tried experimenting with CI_MERGE_REQUEST_LABELS, but it doesn’t seem to be able to read that as an env var from within the script.
You have to use merge request pipelines for the CI_MERGE_REQUEST_LABELS variable (and other MR-related variables) to be present as documented in predefined variables.
You could use a rules: clause to skip jobs. Something like
build:
rules: # only run this job if the regex pattern does not match
- if: $CI_MERGE_REQUEST_LABELS !~ /skip_build/
You can also do this on any other kind of predefined (or user-defined) variable, like branch name, commit messages, MR titles, etc. Whatever works for you.
For example, a built in feature of GitLab is that if your commit message contains [ci skip] it will prevent the pipeline from running. You could implement similar functionality for your jobs and/or pipelines through rules: or workflow:rules:.
I have a GitLab CI pipeline with a 'migration' job which I want to be added only if certain files changed between current commit and master branch, but in my current project I'm forced to use GitLab CI pipelines for push event which complicates things.
Docs on rules:changes clearly states that it will glitch and will not work properly without MR (my case of push event), so that's out of question.
Docs on rules:if states that it only works with env variables. But docs on passing CI/CD variables to another job clearly states that
These variables cannot be used as CI/CD variables to configure a
pipeline, but they can be used in job scripts.
So, now I'm stuck. I can just skip running the job in question overriding the script and checking for file changes, but what I want is not adding the job in question to pipeline in first place.
While you can't add a job alone to a pipeline based on the output of a script, you can add a child pipeline dynamically based on the script output. The method of using rules: with dynamic variables won't work because rules: are evaluated at the time the pipeline is created, as you found in the docs.
However, you can achieve the same effect using dynamic child-pipelines feature. The idea is you dynamically create the YAML for the desired pipeline in a job. That YAML created by your job will be used to create a child pipeline, which your pipeline can depend on.
Sadly, to add/remove a Gitlab job based on variables created from a previous job is not possible for a given pipeline
A way to achieve this is if your break your current pipeline to an upstream and downstream
The upstream will have 2 jobs
The first one will use your script to define a variable
This job will trigger the downstream, passing this variable
Upstream
check_val:
...
script:
... Script imposes the logic with the needed checks
... If true
- echo "MY_CONDITIONAL_VAR=true" >> var.env
... If false
- echo "MY_CONDITIONAL_VAR=false" >> var.env
artifacts:
reports:
dotenv: var.env
trigger_your_original_pipeline:
...
variables:
MY_CONDITIONAL_VAR: "$MY_CONDITIONAL_VAR"
trigger:
project: "project_namespance/project"
The downstream would be your original pipeline
Downstream
...
migration:
...
rules:
- if: '$MY_CONDITIONAL_VAR == "true"'
Now the MY_CONDITIONAL_VAR will be available at the start of the pipeline, so you can impose rules to add or not the migration job
When you work on your .gitlab-ci.yml for a big project, for example having a time consuming testing stage causes a lot of delay. Is there an easy way to disable that stage, as just removing it from the stages definition, will make the YAML invalid from Gitlab's point of view (since there's a defined but unused stage), and in my case results in:
test job: chosen stage does not exist; available stages are .pre, build, deploy, .post
Since YAML does not support block comments, you'd need to comment out every line of the offending stage.
Are there quicker ways?
You could disable all the jobs from your stage using this trick of starting the job name with a dot ('.'). See https://docs.gitlab.com/ee/ci/jobs/index.html#hide-jobs for more details.
.hidden_job:
script:
- run test
There is a way to disable individual jobs (but not stages) like this:
test:
stage: test
when: manual
The jobs are skipped by default, but still can be triggered in the UI:
Also possible with rules and when as below:
test:
stage: test
rules:
- when: never
So far, the easiest way I've found is to use a rules definition like so:
test:
stage: test
rules:
- if: '"1" != "1"'
(...)
This still feels a bit odd, so if you have a better solution, I'll gladly accept another answer.