Apparently there is a problem with the SYSMAN owned dbms_job that executes this OEM procedure. Start by looking to see of the job is broken. You can unbreak it with dbms_job.broken(jobno,FALSE) procedure. If it begins failing and breaking again you will need to determine what is wrong in your configuration.
You may need to shutdown the Management Server(s), connect to the Grid repository and issue the exec emd_maintenance.remove_em_dbms_jobs, exec emd_maintenance.submit_em_dbms_jobs commands if you are getting hung statuses. But start by looking at the dbms_job itself. It may be something very simple like a manually broken job that needs to be unbroken.
This was first published in May 2007