( Correct me if I'm wrong.)
What could be the reason for the fact that the event "1 distinct types of ORA- errors have been found in the alert log." is not clearable?
Even if the problem is solved, the ORA- error will still be in the alert.log.
Therefore the event would eventually only disappear after 7 days.
Doesn't sound very logical to me.
- The event page showing a warning event for "Generic Alert Log Error Status" metric,
- "Generic Alert Log Error Status" event is a statefull metric,
- On each evaluation of the metric defined as 'Statefull', the Cloud Control Agent recalculates the severity,
- This means, EM Agent scans the alert log for ORA errors, "Generic Alert Log Error Status" will return the number of ORA error found,
- Whenever the returned number is grater than the assigned thresholds of "Generic Alert Log Error Status", an alert is raised on EM console,
- On the next scan of alert log file, this alert will be cleared ONLY if the collected value is less than the assigned threshold,
- By checking your system, the warning threshold of "Generic Alert Log Error Status" is 0, so whenever a single ORA error is found in alert log, a warning event will be raised and you will not be able to clear it manually as this is a stateful metric.
There are three cases where this alert can be cleared:
1. The agent found 0 ORA errors in Alert log, then the alert will be cleared automatically
2. Manual clear: By disabling/enabling the metric
3. Manual clear and further not receiving similar alerts frequently: By assigning higher thresholds values
So, I suggest to disable the metric, then, enable it after some time as follows as follows:
- Go to the problematic database home page >> open the 'Oracle Database' drop-down menu >> 'Monitoring' >> 'Metric and Collection Settings',
- Choose 'All metrics' from the 'View' drop-down list >> search for the 'Generic Alert Log Error Status' metric,
- Click the pencil icon under 'Edit' column opposite to the above metric >> remove the Warning Threshold (make it empty) >> 'Continue' >> 'OK'.
- Wait for the next collection schedule in order for the warning events to be cleared, then, enable the metric again with the same steps (setting Warning Threshold=0).
Note 604385.1 Receiving "Clear" Notifications Unexpectedly for 'Generic Alert Log Error Status' Metric
Note 733784.1 What are Statefull and Stateless Metrics in Enterprise Manager - Explanation and Example
Thanks for the reply.
Still I think it's an issue that should be fixed.
It has no use this way.
As far as I know the agent keeps track of the scanned part of the alert.log.
So with every new run of the check it starts from the last scanned row.
Old ORA- errors are not read again.
The thing is:
CC warns me that there is one or more ORA- errors in the alert.log.
I solve or ignore the ORA- errors.
Still I have to wait 7 days (by default) before the warning is cleared.
"Hacking" the alert.log or bounch the metric does not sounds like a nice solution to me.
What is the DB Plugin version in your environment?
What are the warning and critical thresholds for Generic Alert Log Error Status metrics?
After solving the ORA problem, did you clear the corresponding ORA- alert from EM? If not, please clear the alert.
ORA error count is reported based on number of ORA alerts in EM. Count will be reevaluated and count alert will be cleared when user clears the ORA alert from EM and the count falls below the thresholds set by user.
Please verify and let me know if you still see wrong count alert after reviewing this information.
As suggested by Ravi , once you clear the corresponding ORA- alerts from Console the alert should go away
One of the workaround to clear the alert is to re-evaluate the metric with changed threshold
Login to console
Go the "All Metrics" page of an instance -> Click on the event -> Click on the glasses and you will see the Reevaluate
When we click on Re-evaluate Generic Alert Log Generic Status metric, we get following message:
The current setting for this metric requires that its data point be below the warning and critical thresholds for consecutive evaluations before the alert is cleared. If the data point is not below the thresholds, the alert will not be cleared and no new alert will be generated; it will count towards the consecutive evaluations.
Click No to cancel the alert evaluation. The agent will continue collecting data according to its predefined schedule.
Click Yes to temporarily override the metric settings for this evaluation and clear the alert if the metric value is below the thresholds.
Do you want to continue?
We'll have to modify the thresholds based on the alert that we got and click on Reevaluate.
If you see Warning alert something like "1 distinct type of ORA-..." AND thresholds are set 0 for Warning / 1 for Critical, you have to modify the thresholds to be higher the number of distinct type
alert and reevaluate the metric and it should clear the alert
The plugin is 188.8.131.52, I noticed there is a newer one released.
By "corresponding ORA- alert" you mean the specific alert, not the " 1 distinct type of ORA-..." (which is unclearable)?
Yes, those were manually cleared days ago.
Still the "1 distinct type of ORA-" is not clearable and reevaluate also does not do the trick.
It wouldn't make any sense to relate those 2 differtent allerts to eachother this way.
This issue is being investigated. Meanwhile, you can use the following steps to clear the alert for the Generic Alert Log Error Status metric.
- Log on to the host where the target in question resides, and go to directory <AGENT_HOME>/bin.
./emctl config agent listtargets
- Take note of the target name and target type of the target from the output of the above command.
./emctl clearstate agent <target_name>,<target_type>
./emctl control agent runCollection <target_name>:<target_type> db_alertlog_coll_12
- The alert will be cleared shortly.