CI/CD Runner failing with 'Missing Delta Base' Exception

We have implemented a pipeline using:

GitLab 12.8.5 
GitLab Shell11.0.0
GitLab Workhorsev8.21.0
GitLab APIv4
GitLab Pages1.16.0

And an AWS Runner on version 15.8.1, running on Ubuntu.
Recently, when we try to execute the pipeline, the below message is appearing each time:

:prepare (Thread[Daemon worker,5,main]) completed. Took 0.565 secs.
[145]( FAILURE: Build failed with an exception.
[146]( * What went wrong:
[147]( Execution failed for task ':prepare'.
[148]( > org.eclipse.jgit.api.errors.TransportException: **Missing delta base 4cc4be5f642dacedaf5b7bd6e840856355e703a4**
[149]( * Try:
[150]( Run with --debug option to get more log output. Run with --scan to get full insights.

Even, after we rollback our changes, the same issue is still comming up.

Currently, what we have tried with no success:

  • Updating the gradle version to 6.8 and 6.9
  • Clearing the runner’s cache from the GUI
  • Restarting the runner service on the AWS Instance
  • Creating a new branch and triggering a new build from a separate branch
  • Updating the pipeline to match the one from other services that are building successfully.
  • Updated all the dependencies to the latest versions.
  • Using a different runner.

This is the step in which the pipeline fails:

.publish: &publish
    - mkdir -p /root/.gradle
    - export GRADLE_USER_HOME="/root/.gradle"
    - ls -la
  stage: publish
    - ./gradlew clean
    - ./gradlew build --debug --info --stacktrace -Prelease.useLastTag=true 
    - ls
    - ls build/libs
  artifacts: *jars

The ls commands in the before and the after script, are successfully returning the list of files in the folder, and the size is matching with what we can get on our local PC’s.

Also, something that we have noticed, is that the missing delta base hash (4cc4be5f642dacedaf5b7bd6e840856355e703a4) is always the same.

Any ideas of something that we could be missing?