rambrus commented on PR #17789: URL: https://github.com/apache/iceberg/pull/17789#issuecomment-6466477718
@jarosmpost @singhpk234 We're hitting the same problem and would like to help get this merged. Our setup is Spark with Iceberg 1.10.0 and a Glue catalog using `LakeFormationAwsClientFactory`. We measured one `GetTemporaryGlueTableCredentials` call for every S3 GET, a 1:1 ratio. That holds even when the same S3 client is reused across a scan, so at scale the volume of Lake Formation calls gets very large. `LakeFormationAwsClientFactory` is unchanged through 1.12.0, so upgrading doesn't fix it. From the history, it looks like the PR went stale while waiting for re-review after the Sep 4 rework, not because it was rejected. Two questions: 1. @singhpk234, does the rework (CachedSupplier plus a JVM-wide registry keyed by table ARN and catalog properties) address your review comments? Or would you prefer a different direction? 2. @jarosmpost, do you plan to reopen this? If not, would you mind if we picked it up in a new PR and credited you? -- 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]
