Hi there,
we are monitoring our Oracle servers using mk_oracle. Our instances constantly claim that jobs are in a WARN state, like this:
These are the jobs with this specific problem:
FILE_WATCHER
PMO_DEFERRED_GIDX_MAINT_JOB
CLEANUP_NON_EXIST_OBJ
CLEANUP_ONLINE_IND_BUILD
CLEANUP_TAB_IOT_PMO
CLEANUP_TRANSIENT_TYPE
CLEANUP_TRANSIENT_PKG
CLEANUP_ONLINE_PMO
FILE_SIZE_UPD
As far as we can tell, the problem seems to be missing historic data. However, this link to the Oracle support database has this to say about these jobs:
This is by design as these jobs belong to the SCHED$_LOG_ON_ERRORS_CLASS
This job class only logs information on failures, if the jobs are successful, no information is recorded.
The suggested action if historic data is required:
exec dbms_scheduler.set_attribute(‘<JOB_NAME>’, ‘logging_level’,DBMS_SCHEDULER.LOGGING_FULL);
We would suggest either not expecting historic data for jobs or single out the jobs described here from the jobs with expected historic data. The statement to determine which jobs this would entail:
select job_name from dba_Scheduler_jobs where job_class = ‘SCHED$_LOG_ON_ERRORS_CLASS’;