This feature is currently in beta and your organization may not have access to it yet. The functionality and documentation are subject to change.
The Monitoring tab in the Data Collection UI allows you to monitor usage patterns, errors, and compute time of your event forwarding properties. This guide provides a high-level overview of how to view and understand the reports shown in the tab.
This guide assumes that you have purchased event forwarding and that you have a working understanding of how event forwarding works. See the event forwarding overview for more information.
Watch the following video for a high-level overview of the monitoring feature:
Hi, I’m John Viray on the Adobe product team. In this video, I’m going to show how monitoring can help you understand the general usage and endpoint errors in event forwarding properties. By the end of this video, you’ll be able to first, understand the number of incoming and outgoing calls to and from event forwarding. And second, know the number of errors across all properties and even run a query to understand the errors observed from a specific end point, With that, let’s jump into the product. After you select monitoring from the left rail navigation, you’re brought to this screen. Monitoring includes three reports, usage, error events and compute time, The property and environment selector above the usage report controls all three of the reports to make sure the reports are viewed within the same context. The usage report shows a total number of incoming and outgoing calls for a given month. Error events shows the errors observed from endpoints being hit from event forwarding. You can also choose from several different timeframes. Hover over the line graph to see the breakdown of endpoint error response codes and even enter the domain of an endpoint to see how many errors are observed from that particular endpoint. Right now, this query must be an exact match. For example, to see how many errors are reported from the Facebook conversions API, I would enter the end point domain and press Enter. Lastly is compute time. This report does not show end to end latency. It shows the actual time it takes to execute all the rules in the event forwarding property. This compute time excludes any idle time event forwarding spends waiting for an endpoint to respond. We included this data because there is a 50 millisecond limit. However, in all our observations, we haven’t seen any customer come remotely close to this threshold. This wraps up the overview of monitoring and we hope you enjoy the first iteration of this feature. -
You can view metrics within an individual environment and property, or across all properties and environments owned by your organization.
To show metrics for a single property, select the property dropdown menu and choose the property of interest from the list. Once you have chosen a property, you can also use the environment dropdown to select an environment of interest.
The Usage report shows incoming and outgoing calls for a given time period. Incoming calls represent data sent to event forwarding. Outgoing calls represents data sent from event forwarding. The Total events number in the left pane is the sum of incoming and outgoing calls for the given time period.
The Error Events report shows errors in aggregate, and broken out by HTTP response code when you hover your cursor over the line chart. The displayed errors are from outgoing calls and the response codes are from the endpoint that event forwarding is interacting with.
The errors are shown for a given time period, which can be adjusted from the provided dropdown menu.
The search box for the error event allows you to query event forwarding to understand errors for a given endpoint domain. You must enter the exact domain, as the search feature does not accept approximations or “fuzzy” matches. Once you provide an exact domain for which there is outgoing error data, hit Enter and the report refreshes to show outgoing errors for that domain. For example, to see errors from the Facebook Conversions API endpoint the domain should be written as
The Compute Time report shows the compute time of all rules on event forwarding servers.
The displayed times do not represent end-to-end latency. Event forwarding has a compute-time limitation of 50 milliseconds. If this limit is exceeded, the related data will be dropped.
The following factors affect compute time:
For example, if an action in event forwarding hits an endpoint and that endpoint takes two seconds to respond, this two-second latency will not count against compute time because event forwarding is just waiting and not actively computing anything. The response time cannot be longer than 30 seconds, otherwise data will be dropped.