crisfiorini-oss opened a new pull request, #40845: URL: https://github.com/apache/beam/pull/40845
BlobStorageIO.exists currently catches ResourceNotFoundError, but _blob_properties translates that exception into BlobStorageError. A missing blob therefore raises instead of returning False, causing FileBasedSink.pre_finalize / WriteToText to fail for a fresh Azure Blob destination. Catch the translated BlobStorageError and check its code. Preserve failures for every non-404 response. Reuse the existing Azure filesystem and exception translation; no client, retry or pipeline changes. Validation: the new missing-blob regression fails against the unmodified Beam 2.72 SDK with BlobStorageError(None, 404). All eight tests in blobstorageio_test.py (four existing path cases and four existence cases) pass with the fix using Beam 2.72 and the Azure SDK. Coverage includes existing blobs, the real ResourceNotFoundError-to-BlobStorageError translation, authorization failures and non-404 translated failures. Full suites run in CI. The failure was reproduced on Flink 1.20 using the Beam 2.72 portable Python SDK with Azure workload identity; no credentials are needed for the regression tests. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
