datapatch through opatchauto in the last RAC node fails with locks in the database
Applying 19c April'21 PSU (19.11) on Linux x64
2 node RAC cluster
1 node completed successfully
while running opatchauto on the 2nd node, binary patches were successfully applied but the datapatch portion was running for over 20 minutes. When I reviewed session logs, the datapatch was attempting to acquire a lock on an object and since there were live transactions in the database, it failed to acquire lock.
Oracle support suggests to run datapatch after starting database in RESTRICT mode. Well, this is a RAC rolling patch install and I don't understand why should the database be taken down for datapatch? We had similar issues with 19c October'20 PSU and had to take databases down. But can't afford to take an outage every quarter.