Official aws-cli image does not accept scripts commands from gitlab

user_7648881 0
Viewed 326

I am using amazon/aws-cli:2.2.40 in a gitlab pipeline to publish a static site on a S3.

deploy_front: 
  image: amazon/aws-cli:2.2.40
  variables:
    AWS_ACCESS_KEY_ID:  $AWS_ACCESS_KEY_ID
    AWS_SECRET_ACCESS_KEY: $AWS_SECRET_ACCESS_KEY
    AWS_DEFAULT_REGION: $AWS_DEFAULT_REGION
  stage: publish
  script:
    - s3 sync ./front/build s3://some-bucket-name

ends up in

usage: aws [options] <command> <subcommand> [<subcommand> ...] [parameters]
To see help text, you can run:
  aws help
  aws <command> help
  aws <command> <subcommand> help
aws: error: argument command: Invalid choice, valid choices are:
accessanalyzer                           | acm                                     
acm-pca                                  | alexaforbusiness                        
amp                                      | amplify                                 
amplifybackend                           | apigateway
.......                                    ..........

I managed to replicate this error by running docker run -it amazon/aws-cli:2.2.40 aws s3 locally. However docker run -it amazon/aws-cli:2.2.40 s3 works so I do not understand why I get the same error no matter what I use in script. Is it a some escaping issue?

gitlab-ci aws-cli amazon-s3
1 Answers

This is due to the amazon/aws-cli image's entrypoint being: /usr/local/bin/aws. When you run aws by itself locally, it'll return the help message and give a 252 exit code. To override the entrypoint, you do so like this:

test:
  stage: test
  image: 
    name: amazon/aws-cli
    entrypoint: [""]
  script:
    - aws s3 help
user_169782280
Related