The GitHub Actions job "Tests" on airflow.git has failed. Run started by GitHub user potiuk (triggered by potiuk).
Head commit for run: db4689f3c720056fba1c6bbe81decd2e4b1267be / LIU ZHE YOU <[email protected]> Fix k8s flaky test - Improve Backoff Retry (#47811) ## Why Flaky Kubernetes tests have been failing: https://github.com/apache/airflow/actions/runs/13853802423/job/38770312000 All failures occur at `_ ERROR at setup of TestKubernetesPodOperator.test_kubernetes_pod_operator_active_deadline_seconds[10] _`. ( The root cause is still unclear, and there may be other side effects. ) For now, I am fine-tuning the backoff retry configuration to improve stability. ## What Improved the backoff retry configuration to increase retry frequency while maintaining a balanced total retry duration. The previous strategy had long delays between retries and did not provide enough opportunities to recover from transient failures. ### **Comparison of Retry Strategies** | Feature | **New Strategy** (`total=10, backoff_factor=1`) DEFAULT_BACKOFF_MAX = 32 | **Old Strategy** (`total=5, backoff_factor=10`) | |--------------------|------------------------------------------------|----------------------------------| | **Total Retries** | 10 | 5 | | **Backoff Factor** | 1 | 10 | | **Max Wait Time** | **Capped at 32 sec** | No cap (exponential growth) | | **Backoff Formula** | `min(1 × (2^(retry - 1)), 32)` | `10 × (2^(retry - 1))` | | **Backoff Times** | `1, 2, 4, 8, 16, 32, 32, 32, 32, 32` | `10, 20, 40, 80, 160` | | **Total Time** | **191 sec (~3.2 min)** | **310 sec (~5.2 min)** | Report URL: https://github.com/apache/airflow/actions/runs/13881153811 With regards, GitHub Actions via GitBox --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
