Our application is manually reporting user actions with custom string properties appended to each. We have also created a custom metric so that we could make multidimensional charts based on those string properties as dimensions. The multidimensional chart tool as well as USQL are both able to count those user actions and filter based on the string properties. However when we try to create a custom event for alerting based the Static threshold setting for that metric, it only shows time units (of the metric) to trigger the alert; even-though 'Count' is selected under the aggregate dropdown. I'm wondering if this is due to the type of our metric being a 'User action duration' metric. But the RumMetricDefinition object doesnt seem to have anything that could describe a user action count. Is there a way to create an alert that would trigger based on a certain count of custom user actions?
Solved! Go to Solution.
Another way to accomplish this would be to mark the particular user action of interest as a key user action.
Then when you create a custom event, you may filter to this (or several) key user actions using the action name rule-based filter.
This has the static threshold set against the count of the user actions, rather than the performance metric.
Thanks for your response @Andrew M. But it seems like the Metric in your screenshot is "by browser". So I wouldn't be able to split by my custom dimensions) which are the action string properties that i define with custom Metric).
Hi Tim A.,
I was just using a generic example. I believe for your particular case, actually the property can only be applied as a filter when you create the metric. As the property cannot be a dimension (splitting value), then it could not be used in the alerting dimensions selection.
I may be wrong and would need to understand your exact metric and use case in more detail.
I believe you will need to create the metric based on Request Count or similar metric, instead of Response Time. Then only will you get the count and/or sum aggregation that you will need. For response time metric will have the time based aggregations, like average (time), min (time), max (time), etc.