Bash error in ssh runner

Problem to solve

Hi all

I upgraded yesterday from 18.0.x to the latest.
Now my two ssh runners do this:

Saving cache for successful job 00:00
Creating cache yarn-19-protected...
bash: line 168: /home/gitlab-runner/builds/w-q2-u1_r/0/snip/snip/builds/w-q2-u1_r/0/snip/snip.tmp/gitlab_runner_cache_env: No such file or directory

then continues with uploading artifacts with no issue, but at the end ERROR: Job failed: Process exited with status 1, and pipelines fail

Has anyone seen this? We are kinda stuck

Steps to reproduce

I tried recreating the runner, to clean up its home, to downgrade it to 18.3. No luck

Configuration

Something along the lines of

.yarn-cache:
  cache:
    key: yarn
    paths:
      - /home/gitlab-runner/.yarn/berry/cache

.builder:
  extends: .yarn-cache
  variables:
    GIT_SUBMODULE_STRATEGY: recursive
  before_script:
    - source "$HOME/.nvm/nvm.sh"
    - yarn install
    - yarn npm audit --severity moderate
  artifacts:
    paths:
      - dist/
    expire_in: 1 day

.linting: &linting
  - yarn run prettier-check
  - yarn run lint
  - src/guide/check.sh

build-prod:
  extends: .builder
  stage: build
  tags:
    - build_frontend
  script:
    - *linting
    - ./deploy.sh --build
    - >
      ./deploy.sh <snip>
  rules:
    - if: '$CI_PIPELINE_SOURCE == "schedule"'
      when: never
    - if: '$CI_COMMIT_REF_NAME =~ $TODEPLOY'

Versions

Please select whether options apply, and add the version information.

• Self-managed X
• GitLab.com SaaS
• Dedicated
• Self-hosted Runners X

Versions

• GitLab (Web: /help or self-managed system information sudo gitlab-rake gitlab:env:info): v18.4.1-ee
• GitLab Runner, if self-hosted (Web /admin/runners or CLI gitlab-runner --version):18.4.0 both on the gitlab server and the target ssh server