Metadata flush calls does not refresh the cache

[ssivaprathee]

sys flush calls which are submitted to trino are getting completed but the metadata cache in trino is not getting refreshed. This happens intermediately, and this happens only in tables with partitions. Once we refresh the whole the table, the issue is getting resolved

Sample Flush call:

call system.flush_metadata_cache(schema_name=>'schema_name', table_name=>'table_name', partition_columns=>ARRAY['partition_column_name'], partition_values=>ARRAY['partition_column_value'])

[ebyhr]

How did you confirm the cache status?

[ssivaprathee]

After making the flush call with partitions, we tried querying the same table, it was giving stale data. After that I made the flush call without partitions parameter, data was not stale anymore. It happened around 4 to 5 times in the last month.

[ebyhr]

Could you share the entire steps to reproduce with your catalog config properties?

[ssivaprathee]

I am not sure, if u can reproduce as this is intermediate, failing sometimes. Its a hive catalog connecting to AWS Glue.

Fire a flush call for a table with partitions with partitions parameters like mentioned above.. $Path wont get refreshed.
Fire a flush call to the same table without partitions parameters, all $Path will be refreshed