sundapeng opened a new pull request, #10299: URL: https://github.com/apache/paimon/pull/10299
### Purpose Writing a large file to OSS fails as soon as CompleteMultipartUpload is throttled with `503 QpsLimitExceeded`. The Aliyun OSS SDK never retries POST requests other than InitiateMultipartUpload, so CompleteMultipartUpload and DeleteObjects surface the first throttle or 5xx as an error, while GET and PUT on the same client are retried. Under bucket-level throttling this fails Flink writers and lake tiering jobs outright. `OSSFileIO` now installs its own retry strategy on every OSS client. It retries 429, 500, 502, 503, 504 and network errors on all requests, including those POSTs, which are safe to repeat. The backoff stays exponential but is capped at 10s and jittered, so throttled writers do not retry in lockstep. `fs.oss.attempts.maximum` still bounds the attempts. ### Tests `OSSRetryStrategyTest`: 6 new cases, 3 of them against a fake OSS endpoint; 23 tests in `paimon-oss-impl` passed. -- 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]
