You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Random failures (~50% of the time) getting credentials for a cluster. Reason is empty string.
Expected behavior
Request to get credentials should work every time, or never. If it fails, the actions should provide a reason so the issue can be resolved, not reason: with nothing following.
Observed behavior
The action is failing ~50% of the time with no reason to explain the failure.
Action YAML
name: Run Exporter Testson:
pull_request:
types: [assigned, opened, synchronize, reopened]paths-ignore:
- "README.md"workflow_dispatch:
# https://docs.github.com/en/actions/using-workflows/workflow-syntax-for-github-actions#jobsjob_idconcurrencyconcurrency:
group: ${{ github.workflow }}-${{ github.head_ref || github.ref }}cancel-in-progress: truejobs:
run:
runs-on: nscloud-ubuntu-22.04-amd64-2x4timeout-minutes: 20steps:
- uses: actions/checkout@v2
- name: Set up Docker Buildx# This is where the actual error occurs,# It is included in the section for Additional informationuses: ./.github/actions/setup-buildxwith:
credentials_json: "${{ secrets.GCP_GITHUB_ACTION_SA_KEY }}"
- uses: actions/setup-python@v4with:
python-version: "3.11"cache: pip
- name: Setup pythonrun: python -m venv .venv
- name: Install Poetryuses: snok/install-poetry@v1
- name: Install package locallyrun: | source .venv/bin/activate poetry install -C projects/exporter
- name: Install sqitchrun: | docker pull sqitch/sqitch curl -L https://git.io/JJKCn -o sqitch && chmod +x sqitch
- name: Run exporter testsenv:
PROJECT_ID: ${{ secrets.GCP_PROJECT_ID }}run: | source .venv/bin/activate pytest -v projects/exporter/tests -s
Hi @squat - for how long has this issue been occurring? We are returning the raw API response, which appears to be an empty body, so I'm wondering if this is a server-side issue.
Hi @sethvargo, I'm sampling our logs right now and the earliest occurrence I can find is 2024-12-02 17:01 UTC. That said, it was very infrequent until two daya ago. Today it's happening extremely frequently.
Hmm - I'm also confused about the debug log output, since there should be a message that says "Using context name: ...", which runs before any outbound HTTP calls are made. Did you enable step debugging by setting ACTIONS_STEP_DEBUG and ACTIONS_RUNNER_DEBUG to true?
TL;DR
Random failures (~50% of the time) getting credentials for a cluster. Reason is empty string.
Expected behavior
Request to get credentials should work every time, or never. If it fails, the actions should provide a reason so the issue can be resolved, not
reason:
with nothing following.Observed behavior
The action is failing ~50% of the time with no reason to explain the failure.
Action YAML
Log output
Additional information
The text was updated successfully, but these errors were encountered: