In 12c databases, ORA-12805, ORA-07445 [qertqistart] or ORA-600  errors may occur depending on bug 25694206. User operation cannot be completed due to the closing of parallel processes.
The problem is corrected in version 188.8.131.52, and the following workraound can be applied to the problem.
To resolve the problem, the parameter “_px_wif_extend_distribution_keys” must be set to false.
SQL> alter system set "_px_wif_extend_distribution_keys"=false scope=both sid='*';
This solved an issue we were having on a RAC instance on an Oracle Data Appliance…
You’re welcome. Thank you for your comment.
This solution definitely resolved the issue. But I wanted to know as how this parameter is affecting the programs throwing ORA-12805. We were getting this error only from OEM and no error witnessed in alertlog file of the target database nor the em database. Only seen below events in em UI
Wait Events reported at times are
1. resmgr:cpu quantum. (Schedular waits 47.6 %)
2. enq: TX – row lock contention. (causing application waits ~ 33%)