Scheduled job failover behavior in RAC
Doc ID 2365434.1 explains stickiness of scheduled jobs in RAC, and that a correctly configured scheduler job will continue its schedule on a remote instance if the instance on which it was created is down.
After a Data Pump export/import of an application schema from an older version running standalone into 19c RAC, the scheduler jobs are set with INSTANCE_STICKINESS of TRUE but INSTANCE_ID of NULL. Which instance are they sticky to? It is not clear which instance it was created on because impdp ran in parallel across the whole cluster. Unless it counts the instance that impdp was started from?