Say I have a number of jobs that all do similar series of scripts, but need a few variables that change between them:
test a:
stage: test
tags:
- a
interruptible: true
rules:
- if: $CI_PIPELINE_SOURCE == 'merge_request_event'
script:
- echo "env is $(env)"
- echo etcetera
- echo and so on
- docker build -t a -f Dockerfile.a .
test b:
stage: test
tags:
- b
interruptible: true
rules:
- if: $CI_PIPELINE_SOURCE == 'merge_request_event'
script:
- echo "env is $(env)"
- echo etcetera
- echo and so on
- docker build -t b -f Dockerfile.b .
All I need is to be able to define e.g.
- docker build -t ${WHICH} -f Dockerfile.${which} .
If only I could make an anchor like:
.x: &which_ref
- echo "env is $(env)"
- echo etcetera
- echo and so on
- docker build -t $WHICH -f Dockerfile.$WHICH .
And include it there:
test a:
script:
- export WHICH=a
<<: *which_ref
This doesn't work and in a yaml validator I get errors like
Error: YAMLException: cannot merge mappings; the provided source object is unacceptable
I also tried making an anchor that contains some entries under script inside of it:
.x: &which_ref
script:
- echo "env is $(env)"
- echo etcetera
- echo and so on
- docker build -t $WHICH -f Dockerfile.$WHICH .
This means I have to include it from one step higher up. So this does not error, but all this accomplishes is cause the later declared script section to override the first one.
So I'm losing hope. It seems like I will just need to abstract the sections away into their own shell scripts and call them with arguments or whatever.
The YAML merge key << is a non-standard extension for YAML 1.1, which has been superseded by YAML 1.2 about 14 years ago. Usage is discouraged.
The merge key works on mappings, not on sequences. It cannot deep-merge. Thus what you want to do is not possible to implement with it.
Generally, YAML isn't designed to process data, it just loads it. The merge key is an outlier and didn't find its way into the standard for good reasons. You need a pre- or postprocessor to do complex processing, and Gitlab CI doesn't offer anything besides simple variable expension, so you're out of luck.
Related
When running my gitlab ci I need to check whether a specified svn directory exists.
I was using the script:
variables:
DIR_CHECK: "default"
stages:
- setup
- test
- otherDebugJob
.csharp:
only:
changes:
- "**/*.cs"
- "**/*.js"
setup:
script:
- $DIR_CHECK = $(svn ls https://server.fsl.local:port/svn/myco/personal/TestNotReal --depth empty)
- echo $DIR_CHECK
test:
script:
- echo "DIR_CHECK is blank"
- echo $DIR_CHECK
rules:
- if: $DIR_CHECK == ''
otherDebugJob:
script:
- echo "DIR_CHECK is not blank"
- echo $DIR_CHECK
rules:
- if: $DIR_CHECK != ''
the svn command works and echos back the correct reply but $DIR_CHECK does not get set to anything but the original default. It does not store the returned string from the svn command.
How do I store the returned string from an exe in a variable in gitlab ci?
Test run:
Executing "step_script" stage of the job script 00:00 $ $DIR_CHECK =
$(svn ls https://server.fsl.local:port/svn/myco/personal/TestNotReal
--depth empty) svn: E170000: Illegal repository URL https://server.fsl.local:port/svn/myco/personal/TestNotReal' $ echo
$DIR_CHECK Cleaning up file based variables 00:01 Job succeeded
Passing variables between jobs
Unfortunately, you cannot use DIR_CHECK variable the way you described. List of steps to be executed generates before steps actually runs, that means for all of the steps DIR_CHECK will be equal to default. First of all there are few tips how you can pass variables between jobs:
First way
You can add desired command to the before_script section in your .csharp template:
.csharp:
before_script:
- export DIR_CHECK=$(svn ls https://server.fsl.local:port/svn/myco/personal/TestNotReal --depth empty)
and extend other steps with this .csharp.
Second way
You can pass variables between jobs with job artifacts:
setup:
stage: setup
script:
- DIR_CHECK=$(svn ls https://server.fsl.local:port/svn/myco/personal/TestNotReal --depth empty)
- echo "DIR_CHECK=$DIR_CHECK" > dotenv_file
artifacts:
reports:
dotenv:
- dotenv_file
Thirds way
You can trigger or use parent/child pipelines to pass variables into pipelines.
staging:
variables:
DIR_CHECK: "you are awesome, guys!"
stage: deploy
trigger: my/deployment
In the triggered pipeline your variable will exists at the very start moment, and all the rules will be applied correctly.
Solution
In your case, if you really don't want to include otherDebugJob step in your pipeline you can do the following:
First approach
This is quite easy way and this will work, but looks like not a best practice. So, we are already know how to pass our DIR_CHECK variable from setup step , just add some check in the test step script block:
script:
- |
if [ -z "$DIR_CHECK" ]; then
exit 0
fi
- echo "DIR_CHECK is blank"
- echo $DIR_CHECK
Do the almost same thing for the otherDebugJob but check if DIR_CHECK is not empty with if [ -n "$DIR_CHECK" ].
This approach is helpful when your pipeline not contains a lot of steps, but after the test and otherDebugJob follows another few steps.
Second approach
You can fail your setup step and then handle this fail in otherDebugJob step:
setup:
script:
- DIR_CHECK=$(svn ls https://server.fsl.local:port/svn/myco/personal/TestNotReal --depth empty)
- |
if [ -z "$DIR_CHECK" ]; then
exit 1
fi
otherDebugJob:
script:
- echo "DIR_CHECK is not blank"
when: on_failure
This approach is useful if you only want to make some debug stuff after this setup step. After all on_failure jobs, pipeline will be marked as Failed and stopped.
I am trying to deploy a cloud function using cloudbuild.yaml. It works fine if I don't use any conditional statement. I am facing an error when I execute my cloudbuild.yaml file with if conditional statement. What is the correct way to write it. Below is my code:
steps:
- name: 'gcr.io/cloud-builders/gcloud'
id: deploy
args:
- '-c'
- 'if [ $BRANCH_NAME != "xoxoxoxox" ]
then
[
'functions', 'deploy', 'groups',
'--region=us-central1',
'--source=.',
'--trigger-http',
'--runtime=nodejs8',
'--entry-point=App',
'--allow-unauthenticated',
'--service-account=xoxoxoxox#appspot.gserviceaccount.com'
]
fi'
dir: 'API/groups'
Where am I doing it wrong ?
From the github page, https://github.com/GoogleCloudPlatform/cloud-sdk-docker, the entrypoint is not set to gcloud. So you cannot specify the arguments like that.
Good practice for specifying directory is to start with /workspace
Also the right way to write the step should be
steps:
- name: 'gcr.io/cloud-builders/gcloud'
id: deploy
dir: '/workspace/API/groups'
entrypoint: bash
args:
- '-c'
- |
if [ $BRANCH_NAME != "xoxoxoxox" ]
then
gcloud functions deploy groups
--region=us-central1
--source=.
--trigger-http
--runtime=nodejs8
--entry-point=App
--allow-unauthenticated
--service-account=xoxoxoxox#appspot.gserviceaccount.com
fi
I'm not sure you can do this.
In my case, I use the branch selector in the Cloud build trigger to select which branch (or tag) I want to build from a pattern.
I wanted to delete the latest version of each service only if there were more than two previous versions. This was my solution.
args:
- "-c"
- |
if [[ $(gcloud app versions list --format="value(version.id)" --service=MY-SERVICE | wc -l) -ge 2 ]];
then
gcloud app versions list --format="value(version.id)" --sort-by="~version.createTime" --service=admin | tail -n -1 | xargs gcloud app versions delete --service=MY-SERVICE --quiet;
fi
I am trying to trigger a particular job in CI after either of the 2 conditions
trigger by another job in the same pipeline
OR
changes: somefile.txt
My CI is as described
job1:
stage: build
script:
- echo "JOb1"
- curl -X POST -F token=2342344444 -F "variables[TRIGGER_JOB]=job1" -F ref=master https://main.gitlab.myconmpanyxyz.com/api/v4/projects/1234/trigger/pipeline
only:
changes:
- job1.md
job2: # This does not RUN as expected because of the TRIGGER_JOB set to job1
stage: test
script:
- echo "Job2"
rules:
- if: $TRIGGER_JOB =="job2"
job3: # this RUNS as expected because of VARIABLE TRIGGER_JOB
stage: test
script:
- echo "Job3"
rules:
- if: $TRIGGER_JOB =="job1"
job4: # this also RUNS, but this should not be the expected behavior
stage: test
script:
- echo “job4“
rules:
- if: $TRIGGER_JOB == "xyz"
- changes:
- job4.md
After job1 finishes it also needs to call job4 and not any other jobs (job2 in this case). So I am using CURL to call the job itself. If there are any better ways of calling a specific job in the same CI, also please let me know.
I have already seen this stack-overflow page, but it does not help because my job needs to be triggered by either of 2 conditions which is not allowed bit gitlab-ci.
I need job4 to be called by either of the 2 conditions - if the TRIGGER_JOB=="job1" or if there are any changes in job4.md file.
Currently job4 runs if changes are made in job4.md file, however it also runs if the job1 is triggered. But afaik this should not be the expected behavior.
docs. Can anyone please give me some leads how to create this kind of design.
Your solution was almost correct, but the changes keyword with only or except does only work, if the pipeline is triggered by a push or a merge_request event. This is defined in the variable CI_PIPELINE_SOURCE. When you trigger the pipeline by calling the API, the variable CI_PIPELINE_SOURCE contains the value trigger and therefore only:changes returns always true, which triggers job1 again and ends in an endless loop. You can add a simple except rule to your job1 to prevent that:
job1:
stage: build
script:
- echo "JOb1"
- curl -X POST -F token=2342344444 -F "variables[TRIGGER_JOB]=job1" -F ref=master https://main.gitlab.myconmpanyxyz.com/api/v4/projects/1234/trigger/pipeline
only:
changes:
- job1.md
except:
variables:
- $CI_PIPELINE_SOURCE == "trigger"
You can find more information on only/except:changes in the documentation.
I have an unity ci-project.
.gitlab-ci.yml contains base .build job with one script command. Also I have multiple specified jobs for build each platform which extended base .build. I want to execute some platform-specific commands for android, so I have created separated job generate-android-apk. But if it's failing the pipeline will be failed too.(I know about allow_failure). Is it possible to extend script section between jobs without copy-pasting?
UPDATE:
since gitlab 13.9 it is possible to use !reference tags from other jobs or "templates" (which are commented jobs - using dot as prefix)
actual_job:
script:
- echo doing something
.template_job:
after_script:
- echo done with something
job_using_references_from_other_jobs:
script:
- !reference [actual_job, script]
after_script:
- !reference [.template_job, after_script]
Thanks to #amine-zaine for the update
FIRST APPROACH:
You can achieve modular script sections by utilizing 'literal blocks' (using |) like so:
.template1: &template1 |
echo install
.template2: &template2 |
echo bundle
testJob:
script:
- *template1
- *template2
See Source
ANOTHER SOLUTION:
Since GitLab 11.3 it is possible to use extend which could also work for you.
.template:
script: echo test template
stage: testStage
only:
refs:
- branches
rspec:
extends: .template1
after_script:
- echo test job
only:
variables:
- $TestVar
See Docs
More Examples
We are facing a problem where we need to run one specific job in gitlab CI. We currently not know how to solve this problem. We have multitple jobs defined in our .gitlab-ci.yml but we only need to run a single job within our pipelines. How could we just run one job e.g. job1 or job2? We can't use tags or branches as a software switch in our environment.
.gitlab-ci.yml:
before_script:
- docker info
job1:
script:
- do something
job2:
script:
- do something
You can use a gitlab variable expression with only/except like below and then pass the variable into the pipeline execution as needed.
This example defaults to running both jobs, but if passed 'true' for "firstJobOnly" it only runs the first job.
Old Approach -- (still valid as of gitlab 13.8) - only/except
variables:
firstJobOnly: 'false'
before_script:
- docker info
job1:
script:
- do something
job2:
script:
- do something
except:
variables:
- $firstJobOnly =~ /true/i
Updated Approach - rules
While the above still works, the best way to accomplish this now would be using the rules syntax. A simple example similar to my original reply is below.
If you explore the options in the rules syntax, depending on the specific project constraints there are many ways this could be achieved.
variables:
firstJobOnly: 'false'
job1:
script:
- do something
job2:
script:
- do something
rules:
- if: '$firstJobOnly == "true"'
when: never
- when: always
We faced the same problem in the past and I'm sharing with you our solution.
#Remark#
I read the answer of Jawad and I found it good and we have tried it when we faced the issue.
My remark is that adding when: manual will always show ALL your jobs in the pipeline.
So if you work in a large team, you can't prevent other collaborators to click by error or by mistake on the job you don't want to be launched.
#What I'm supposing before continuing#
Let's say that you have 4 jobs.
You need to always run (manually or automatically) job 1, job 2 and job 4 but NOT job3.
You want to only run job 3 in a specific case or just when you decide to run it.
#The idea is#
We launch the 3rd job only for tags which match a regular expression.
In the example below, it's launched for tags like helloTag.1, helloTag.2, helloTag.3... etc.
If we are in develop or master (or other branch), we will have 3 stages (stage 1, stage 2, stage 4)
Note how the 3rd job is not present in the pipeline
Go to "Repository" --> "Tags" --> "New tag"
Give the tag a name which much your regular expression
If we are in a tag having a name which starts with "helloTag.", we will have 1 stage (stage 3)
Note how other stages are not present here
#Example of .gitlab-ci file#
stages:
- myStage1
- myStage2
- myStage3
- myStage4
This is my first stage:
stage: myStage1
before_script:
- echo "my stage 1 before script"
script:
- echo "my stage 1 script"
except:
- /^helloTag.*$/
This is my second stage:
stage: myStage2
before_script:
- echo "my stage 2 before script"
script:
- echo "my stage 2 script"
except:
- /^helloTag.*$/
This is my third stage:
stage: myStage3
before_script:
- echo "my stage 3 before script"
script:
- echo "my stage 3 script"
only:
- /^helloTag.*$/
This is my fourth stage:
stage: myStage4
before_script:
- echo "my stage 4 before script"
script:
- echo "my stage 4 script"
except:
- /^helloTag.*$/
Hope that this helps you.
Simply add a when: manual to the jobs you don't want to run.
These jobs will still appear in your pipeline but won't be run, unless someone "manually" starts them through the web interface, hence the name.
Here's more info about this: https://docs.gitlab.com/ce/ci/yaml/README.html#when
If you're looking for something more "programmable", let's say run either job1 or job2 depending on a branch name or a tag, then you should have a look at the only and except keywords: https://docs.gitlab.com/ce/ci/yaml/README.html#only-and-except
> Currently it seems not to be possible with GitLab CI to have other software switches than tags or branches as provided in the other answers.
We finally switched to an other "real" CI due to too many limitations on GitLab CI. GitLab CI is unfelixble if you want to run some custom jobs in different procedures. I realy appreciated the both answers here. I'm sure they will help other users to manage this stuff. Unfortunately in our case we could not use tags, commit messages or branches as a software switch.
We are still looking for a answer on this. Feel free to give an other approach to solve this problem. I will mark the right answer once it hits. Also a bounty on this question did not result in an right answer.
The original question asks how to trigger jobs without using branch-names or tags. This leaves commit messages and environment variables as viable sources, and neither require editing your yaml file for each push.
Commit Messages
Job rules with commit message regex is a very simple and flexible solution, in my experience.
Set up your .gitlab-ci.yml like this
job1:
...
rules:
- if: '
$CI_COMMIT_MESSAGE =~ /.*run job1.*/ ||
$CI_COMMIT_MESSAGE =~ /.*run all.*/
'
job2:
...
rules:
- if: '
$CI_COMMIT_MESSAGE =~ /.*run job2.*/ ||
$CI_COMMIT_MESSAGE =~ /.*run all.*/
'
Push with a commit message like this
git commit --allow-empty -m "testing conditional job triggers for gitlab-ci based on branch names and commit messages. run job1"
git push
You'll notice this also allows you to run both jobs with a message like this
git commit --allow-empty -m "testing all jobs. run all"
git push
Environment Variables
Elsewhere in your answer to your own question you add a constraint: commit messages cannot be used. Environment variables can be set in the git cli
Set up your .gitlab-ci.yml like this
job1:
...
rules:
- if: $JOB1
job2:
...
rules:
- if: $JOB2
Push like this
git commit --allow-empty -m "triggering job1 with ci variables"
git push -o ci.variable="JOB1=anythingAtAll"
reference: https://docs.gitlab.com/ee/user/project/push_options.html