Oracle Database 220.127.116.11 on Oracle Enterprise Linux 6.
Import (IMPDP) job on a RAC database failed with very little error detail. The import executed for about 3 hours before failing.
Job "SYS"."SYS_IMPORT_FULL_01" stopped due to fatal error at Sat Feb 8 18:28:09 2020 elapsed 0 02:57:42
Checking the database alert log gave details about the error.
Sat Feb 08 18:29:22 2020 DW0O terminating with fatal err=39078, pid=354, wid=25, job SYS.SYS_IMPORT_FULL_01
> oerr ora 39078 39078, 00000, "unable to dequeue message for agent %s from queue \"%s\"" // *Cause: The Data Pump's communication layer was unable to retrieve a // message from the control or status queue. Subsequent messages will // detail the problem. // *Action: Fix the problem if possible, or contact Oracle Customer Support.
I have not seen this error before. There was no subsequent message to detail the problem. The error clarifies that the problem was with Data Pump’s communication layer and it involves the queue. My import job was running across RAC instances and I was using parallel 24. Metalink note 2318625.1 details the problem.
1 Reports ORA-29913/ORA-00448/ORA-39078 In RAC Environment (Doc ID 2318625.1) (opens in a new tab)”>DataPump Job With Parallel>1 Reports ORA-29913/ORA-00448/ORA-39078 In RAC Environment (Doc ID 2318625.1)
This note applies to 18.104.22.168 and it mentions the issue happens only when an instance is in mounted mode. Neither is my case. I run 22.214.171.124 and my instances were open ‘READ WRITE’.
Since my situation does involve RAC and PARALLEL processing, I decided to use Solution 3 from the note. That fixed the problem.