2018-02-15 11:45 AM
The current mechanism for excluding resources from alerting in OCUM 7.3 doesn't make a whole lot of sense.
If I want to include a cluster in monitoring but not alert on anything for it, the only way to do so is to find and exclude EVERY node, lif, volume, aggregate, storage shelf, etc. Even when I think I've excluded everything, somehow some alert still ends up coming through and paging out.
You would think if you selected the top level cluster to be excluded it would automatically exclude all child objects of the cluster.
2018-02-16 12:50 AM
I have exactly the same issue. I thought I did something wrong. And I thought to ask support for help.
I tried to exclude SVM DR vservers from alerting (they down by design, but OCUM doesn't know about it)
2018-02-16 11:04 AM
While this is definitely a good place for Unified Manager to tighten up the Alerting feature, another way to EXCLUDE alerting from a particular cluster or a particular object, or a particular event on a particular object, is to not to INCLUDE it in the first place.
These are two different workflows.
EXCLUDE: Configuration\Manage Events
For example, if you are getting events from a cluster that you don't want alerts, that means the cluster is already in the first list.
In order to avoid this, when you create the alert, don't use the <<All [object]>> option. Add only the clusters and objects you want alerted on. -- Again, you have to remember not to include other objects, such as node, volumes, LUNs etc.. from the list. Include or exclude cluster does not automatically include or exclude the objects underneath. This is where UM could improve.
The other way is of course what you have tried - exclude the particular cluster, object or event after it has already been included. In this case, you already know exactly what not to alert on. Use information from the alert to trace back, then shut down the flood gate.
Hope this helps (or at least did not confuse you even more ),
2018-02-16 11:22 AM
Thanks Julia for the thoughts!
We've definitely considered that approach. However the problem with NOT including all clusters,volumes,etc and adding in everything manually, is that anytime a new volume is created, a new lif is added, etc. it will not automatically be included in paging. Our environment is huge and constantly changing so this really isn't feasible.