ayushtkn opened a new pull request, #11875: URL: https://github.com/apache/ozone/pull/11875
## What changes were proposed in this pull request? `S3MultipartUploadCommitPartRequest.validateAndUpdateCache` publishes the new part to the cache — `multipartPartsTable`, or in the legacy layout the `multipartInfoTable` entry that carries the part inline — and puts a delete marker on the part's open key, all before calling `checkBucketQuotaInBytes`. When that check throws `QUOTA_EXCEEDED` the response writes nothing and nothing withdraws those entries, so they stay visible until the double buffer has flushed the transaction. `applyTransaction` does not wait for that flush, so the next request for the same upload can be applied while the rejected part still looks committed: an abort or complete reads the parts through the cache and releases bytes that were never charged or completes a key over a part whose commit failed, and in the legacy layout a later part commit starts from the cached entry and writes the rejected part back to `RocksDB`. This PR moves the three cache writes below the quota check, next to the bucket usage update that is already published after it, so a rejected commit leaves no cache state behind. ## What is the link to the Apache JIRA https://issues.apache.org/jira/browse/HDDS-16787 ## How was this patch tested? UT -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
