During development, I drop table table_frq in SQL query mannually. Then run a python notebook using serverless compute, use spark.catalog.tableExists(table_frq) as the condition. Last week, after drop table, spark.catalog.tableExists(table_frq) showed "false". However, yesterday it showed "true". I tried serverless and classic compute using RunTime 17.3ML LTS and 18.0ML, all showed "true". Is there anything changed in #UC? The suspicion is the table's metadata corrupted. But how

Then today "drop table" works again using SQL query. I am concerned about not knowing the reasons for the change and the system being not stable. If anyone could share your experience it is greatly appreciated.