Intermittent error: Docker Login failed with exit code 1 running jobs in container

Describe the bug I am running a job in a container using self-hosted runners and sometimes an error related to docker login occurs so we need to re-run the job again and everything works fine.

I know it can be related to my credentials but I have another job running aws ecr get-login-password command to get the token before this job below.

Note: I am using a private ECR.

My job:

job_name:
    name: Job Name
    runs-on: self-hosted
    container:
      image: 12345678910.dkr.ecr.eu-west-1.amazonaws.com/ecr_name:tag
      credentials:
        username: AWS
        password:  ${{ secrets.password }}

Expected behavior

Run job without docker login error

Runner Version and Platform

EKS version: v1.22.9-eks-a64ea69 Current runner version: ‘2.293.0’

What’s not working?

[rsilva-rs]

@rlinstorres Adding to the proposed fix by @ksdc-andreo you may be facing an intermittent escaping or special char mismatch. Have you tried the following?

- run: echo `(aws ecr get-login-password --region $AWS_REGION)` | gh secret set password

[tom-robo]

However, this does mean that I can’t start a job inside a container (I can only launch containers, etc. as steps within a job).