Is using Beam SQL just massively more resource-intensive or is there a bug somewhere here (in my code or elsewhere)? (I'm using Flink runner)
Here is my code: https://pastebin.com/nNtc9ZaG Here is the error I get (truncated at the end because it's so long and seemingly repetitive) when I run the SQL transform and my memory/CPU usage skyrockets: https://pastebin.com/mywmkCQi For example, After several early firing triggers, 13-15% CPU, 1.5GB-2GB RAM, everything working fine: `rowStream.apply(Combine.globally(Count.<Row>combineFn()).withoutDefaults()).apply(myPrint());` After a single early firing trigger, CPU usage shoots to 90%+, 4.7GB+ RAM, soon crashes: `rowStream.apply(SqlTransform.query("SELECT COUNT(*) FROM PCOLLECTION")).apply(myPrint());` I can't imagine this is expected behavior but maybe I'm just ignorant of how SQL is implemented. Most of this code is just getting Schema Registry Avro Kafka messages into a Row stream. There have been discussions on the mailing list recently about how to do that. This is the best I could do. If that part is incorrect I'd be glad to know. Any help appreciated. Thank you.
