APPLIES TO:

SYMPTOMS

After upgrade to 11.2.0.3, we see wait event resmgr:pq queued permanently.

CAUSE

 Not enough parallel slaves available.

SOLUTION

Increased parallel_servers_target parameter.  Parallel_servers_target should be 50-75% of parallel_max_servers.   Recent experience has shown that fewer DOP downgrades occur when it's 50% of parallel_max_servers.  This is because complex SQL statements can use >  2 slave sets.