SkyPilot: Run AI and batch jobs on any infra (Kubernetes or 12+ clouds). Get unified execution, cost savings, and high GPU availability via a simple interface.
The error message suggests a cloud inference issue:
ValueError: Cannot infer cloud from (region 'us-east-2', zone None). Multiple enabled clouds have region/zone of the same names: [AWS, Lambda]. To fix: explicitly specify `cloud`.
However, this appears to be a resource availability issue causing the test to be flaky, rather than an actual cloud inference problem. The test sometimes passes and sometimes fails, indicating a potential race condition or resource constraint in the test environment.
The test
test_infer_cloud_from_region_or_zone
intest_optimizer_dryruns.py
occasionally fails in GitHub Actions. For example: https://github.com/skypilot-org/skypilot/actions/runs/11737074167/job/32697299221The error message suggests a cloud inference issue:
However, this appears to be a resource availability issue causing the test to be flaky, rather than an actual cloud inference problem. The test sometimes passes and sometimes fails, indicating a potential race condition or resource constraint in the test environment.
We should either:
pytest-rerunfailures