I have a Gitlab CI runner running on windows 10:
before_script:
- "echo off"
- 'call "%VS120COMNTOOLS%\vsvars32.bat"'
- echo.
- set
- echo.
stages:
- build
build:
stage: build
script:
- 'StatusTest.exe'
#- msbuild...
I am trying to fail the build with StatusText.exe (I tried returning status codes -1,0,1; throwing an exception, etc.) But Runner only logs the exception and continues with following steps.
What determines that CI shell runner should fail the build and not proceed to next step?
Output:
...
windows_tracing_logfile=C:\BVTBin\Tests\installpackage\csilogfile.log
$ echo.
$ StatusTest.exe
Unhandled Exception: System.Exception: tralala
at StatusTest.Program.Main(String[] args)
$ echo "Restoring NuGet Packages..."
...
What determines that CI shell runner should fail the build and not proceed to next step?
If a pipeline job exits with the code other than 0
then that job fails causing all the following jobs in the pipeline to be skipped.
This behaviour can be changed on a per job basis with allow_failure job keyword.
To make a job to fail forcefully you need to artificially exit from a job with code other than 0
. Here is an gitlab-ci.yml
job example :
some-example-job:
script:
- # ....
- exit 1
See the GitLab CI UI sreeenshot example. The third job has failed.
On the opposite remove exit 0
and your job would succeed if the remaining script
section commands do not exit with code other than 0
.
Now see all the jobs & the entire pipeline finished successfully.