Hide Forgot
Steps to Reproduce: See the attached Model Project Set. - Transformation SQL has a TTL of 30000 - Add a row to the source table - Query the transformation a couple of times then delete the added row - Wait > 30 seconds - Query again (should be same row count) - Query again and it should be the row count after the row was deleted. project_key: SOA With the following transformation SQL, regardless of the time allowed to elapse, the cache does not expire/invalidate. I'm running the production profile. Here's the transformation SQL text: /*+ cache(ttl:30000) */ SELECT PartsSourceA.SUPPLIER_PARTS.SUPPLIER_ID, PartsSourceA.SUPPLIER_PARTS.PART_ID, PartsSourceA.SUPPLIER_PARTS.QUANTITY, PartsSourceA.SUPPLIER_PARTS.SHIPPER_ID, PartsSourceB.SUPPLIER.SUPPLIER_NAME, PartsSourceB.SUPPLIER.SUPPLIER_STATUS, PartsSourceB.SUPPLIER.SUPPLIER_CITY, PartsSourceB.SUPPLIER.SUPPLIER_STATE, CONCAT2(PartsSourceB.SUPPLIER.SUPPLIER_CITY, CONCAT2(', ', PartsSourceB.SUPPLIER.SUPPLIER_STATE)) AS City_State FROM PartsSourceA.SUPPLIER_PARTS, PartsSourceB.SUPPLIER WHERE PartsSourceA.SUPPLIER_PARTS.SUPPLIER_ID = PartsSourceB.SUPPLIER.SUPPLIER_ID It's not expiring/invalidating the cache unless I CALL SYSADMIN.refreshMatView(viewname=>'PartsVirtual.SupplierInfo', invalidate=>true)
Attachment: Added: PartsForMatViewTesting_MPS.zip
I also see the command logging shows: "Loaded materialized view table #MAT_PARTSVIRTUAL.SUPPLIERINFO from cached contents from another clustered node." This would be a whole lot cooler if there was another cluster member. This is not a cluster. Never was. Possibly the next Jira??
Link: Added: This issue Cloned to SOA-3306
Tested and closed!