This discussion has been locked. The information referenced herein may be inaccurate due to age, software updates, or external references.
You can no longer post new replies to this discussion. If you have a similar question you can start a new discussion in this forum.

Appinsight for SQL: Filter and include component INSTANCE in alert

Appinsight for SQL monitors many application components. Some components cover many instances.

Is it possible:

A: To filter out in the alert "Alert me when a component goes into warning or critical state" certain instances.

For example component "Top indexes with fragmentation" is permanently in critical state because some indexes get fragmentened
very rapidly. I would like to filter these indexes out in the alert and only alert when a "real" index goes bad..
The offending index should still be marked critical, it just does not count in the alert.

B: To include the compinent instance in the alert text.

"Top indexes with fragmentation on index primary_ix of table abcd is critical"

"Top indexes with fragmentation on index ${component.instance.name} of table ${component.instance.parent.name} is critical"

  • You can edit the template to exclude indexes below a certain size, which i usually set to a much higher threshold than they come out of the box because there is not really much performance impact to fragmented small indexes anyway.  I find that once you weed out all the ones that arent at least maybe 50mb or more then the list begins to become semi useful.  Taking it a step further, if you know you have all your defrag jobs scheduled and running in SQL tasks then is it even necessary to set a threshold within Orion on them at all?  It is an extremely odd situation where a DBA would decide that they have their defrag job scheduled tonight but things are looking sketchy this afternoon so we're going to defrag immediately.  If we have good processes in place I just remove the index fragmentation threshold completely.

    To your B, it'd require a custom SQL/SWQL variable, but yes it is doable to get the name of the index that triggered the alert.