On deleting a column from hudi table, its still present when querying from presto using hive connector about presto HOT 3 OPEN

@codope

from presto.

@sutodi Accessing Hudi tables using Presto's hive connector relies on the metadata present in the metastore. If the column still exists in the metastore, you would see it in the describe table output or if you do a SELECT * query. I can see there was an error when you tried to do drop column in spark, so it looks like the metadata in the metastore still has the dropped column. The reason you don't see the issue when querying from spark is probably because spark is reading the metadata from the hudi metadata in the filesystem.

Can you verify whether the column still exists in the metastore?

from presto.

@sutodi You can also try out the Presto Hudi Connector and see if you face the same issue. Ideally, you should see similar behaviour as spark when querying via the Presto Hudi Connector since it also relies on the metadata present in the filesystem. There are different ways to access your Hudi data when using the Hudi Connector. You can have a look at the docs for getting started: https://prestodb.github.io/docs/current/connector/hudi.html

from presto.