Pages

Sunday, June 26, 2016

Concurrent Processing - No Concurrent Manager is Defined to Process This Request on Shutdown of Managers

Concurrent Processing - No Concurrent Manager is Defined to Process This Request on Shutdown of Managers

Error:

When shutting down the concurrent manager with adcmctl.cmd, the Abort Concurrent Manager request goes into Pending Standby status.  Checking the request details for the Abort Concurrent Manager request, shows a warning message as detailed below:

"No concurrent manager is defined to process this request"

Cause:

The issue was caused as time given for all FNDLIBR processes to shutdown gracefully was insufficient.

The Abort Concurrent Manager request completes successfully after a couple of minutes.
When the request details were checked immediately after adcmctl.cmd was executed, the warning
message 'No concurrent manager is defined to process this request' did appear and request was in Pending
Standby status.

When checking the request details again, it showed that the request completed Normal after waiting a few more minutes to view the details.

Solution:

To implement the solution, please execute the following steps:

1. Stop managers (adcmctl.cmd)

2. Open Windows Task Manager to view processes

3. Open View Concurrent Requests form and refresh screen often and wait for
the Abort Concurrent Manager request to "Complete" "Normal".

4. Refresh the Task Manager to verify FNDLIBR and all other concurrent manager
processes are gone.

5. For routine maintenance, use Note.134007.1 CMCLEAN.SQL - Non Destructive Script to Clean Concurrent Manager Tables to clean the tables while the managers are down and then restart managers.

6.  Check managers for requests in Pending status via View Concurrent Requests.

7.  If after cmclean.sql, pending requests remain, the requests must be manually cancelled.  This does not  require shutting down of the managers.  Perform the following in SQL*Plus as the APPS user:

    UPDATE  fnd_concurrent_requests
    SET     phase_code = 'C', status_code = 'E'
    WHERE request_id = '<request_id>';
    COMMIT;

8.  After manually killing request, the job(s) go to Completed Error status and managers should start gracefully without any pending requests.

Concurrent manager shutdown can take a few minutes to complete, wait a few minutes and confirm that it shuts down.  Concurrent Processing development team have confirmed that the concurrent manager does not terminate OS/Database processes, this must be done through some other mechanism such as DCD (Dead Connection Detection via Sqlnet) or some OS method.

Reference metalink Doc ID 342932.1