Fix to avoid retrying the metadata loading once a FileNotFoundException is thrown, when trying to drop Iceberg tables, whose metadata is deleted from its S3 storage

[dmariamgeorge]

Description

According to the current behaviour, when a table's metadata is deleted from S3, it repeatedly tries to access the metadata file, leading to multiple retries and delays in dropping the table. with this fix, upon encountering a FileNotFoundException, it immediately halts retries and proceeds with the drop operation, completing the task in a fraction of the time.

Motivation and Context

Fixes issue: #23511

Impact

Users will experience significantly faster drop operations for Iceberg tables whose metadata has already been deleted from S3 storage. Instead of encountering long delays due to unnecessary retries, it will now recognize when the metadata is missing and complete the drop operation more quickly.

Test Plan

1. Connected Presto to an S3 Minio storage through the Iceberg connector.
2. Created a schema, and within the schema created a table.
3. Now, deleted the metadata and the bucket.
4. Tried dropping the associated table.
5. The table drop operation got completed quickly.

Contributor checklist

• Please make sure your submission complies with our development, formatting, commit message, and attribution guidelines.
• PR description addresses the issue accurately and concisely. If the change is non-trivial, a GitHub Issue is referenced.
• Documented new properties (with its default value), SQL syntax, functions, or other functionality.
• If release notes are required, they follow the release notes guidelines.
• Adequate tests were added if applicable.
• CI passed.

Release Notes

== RELEASE NOTES ==

Iceberg Connector Changes
* Fix to reduce drop time for Iceberg tables with deleted metadata in S3 storage. :pr:`23510`

[tdcmeehan]

Nice change @dmariamgeorge . Can you please try to create a test case for this?

[dmariamgeorge]

Nice change @dmariamgeorge . Can you please try to create a test case for this?

@tdcmeehan - There exists a testcase testTableDropWithMissingMetadata() for this:

presto/presto-iceberg/src/test/java/com/facebook/presto/iceberg/TestIcebergMetadataListing.java

Line 124 in d7ca7af

public void testTableDropWithMissingMetadata()

[hantangwangd]

Nice catch, the change looks good to me!

Maybe in TestIcebergMetadataListing we can set iceberg.hive.table-refresh.max-retry-time to a longer period, and validate the timeOut of testTableDropWithMissingMetadata() to a shorter period, so that we can confirm that it did not trigger the retry?

[steveburnett]

Nit of formatting on the release note entry - need a space before :pr:

== RELEASE NOTES ==

Iceberg Connector Changes
* Fix to reduce drop time for Iceberg tables with deleted metadata in S3 storage. :pr:`23510`

[hantangwangd]

Thanks for the fix.