Description of problem:
Every execution of a refresh pools job for organization 11551121 fails in production due to a null pointer exception (see https://docs.google.com/document/d/1Ka2Ct1MYA8fleSLlI8CnyZVeV7DrFFTbme5lU5J1azs/edit?usp=sharing)
The root cause on most stack traces is elided, possibly because of a JVM optimization. The root cause is included in older stack traces on splunk and appears to indicate that a derived product or its ID is null for either the master pool or an existing pool, as noted in the document above.
Find the stack trace on 2020-AUG-18 in splunk: `candlepin_hosts(env=prod)` job=8a85f99173e8810e0173ee37a4a03d4a | transaction job
Version-Release number of selected component (if applicable): 3.1.16
How reproducible: Always for org 11551121, but reason is unknown.
Note also that this is associated with INC: https://redhat.service-now.com/surl.do?n=INC1397105