Set up alerts that let you know when invocations start to fail, so you can address any issues as soon as possible.

1. Log in to your MessageBird Dashboard.
2. On the navigation panel, click Flow Builder.
3. Look for the flow that you want to turn notifications on for, and click on the bell icon. You can also turn alerts on from a flow's logs/insights page. A pop-up will appear.
4. Toggle the button to Alerts ON.
5. Under Time range, choose the time range that you want to monitor from the drop-down menu. For example, if you select a time range of the 'Last 30 minutes', the alert will analyze all the invocations that happened over the last 30 minutes and decide whether or not an alert needs to sent out.
6. Under Failure threshold, look for the first checkbox and choose what percentage of failed invocations you want to trigger the alert. By default, this is set to 1%. A 1% failure rate is a good starting point, and you can always adjust this at a later date.
   Configure this first checkbox when you expect the number of invocations to fluctuate, and you want to see a dynamic notification threshold. For example, if you set this to 1%, you'll get an alert when 1% of all the invocations you had in the last 30 minutes were errors, regardless of whether there were 100 invocations or 1000 invocations.
   
7. Under Failure threshold, look for the second checkbox and choose the absolute number of errors that you want to trigger an alert. 
   Configure this second checkbox when you want to set a minimum threshold for your alerting service. For example, if you set this to 5, you'll get an alert when 5 of the invocations you had in the last 30 minutes were errors, regardless of if there were 5 invocations or 1000 invocations.
   
8. You can use the two checkboxes in unison to reduce 'false alerts'. For example, if you set the Time range to 5 minutes, the percentage or 'relative value' Failure threshold (the first checkbox) to 1%, and the failure amount or 'relative value' Failure threshold (the second checkbox) to 5, then you'll be notified if there were more than 1% failures in the last 5 minutes, but only if that is more than 5 failures in total. This will prevent you from receiving alerts when you only have a low number of invocations, with every failure that happens or if you have a large number of invocations and only a small % (but still more than 5) fail.
   
9. Depending on your specific configurations, decide if you want to use the And operator or the Or operator. And means that both conditions must be true, and Or means that at least one condition must be true
10. Under Alert flow, the flow that will send you alerts is pre-defined. If you're using this for the first time, you'll need to specify which email address you want the alerts to be sent to.
    
11. Once you've configured your alert, take a look at the Summary at the bottom of the pop-up. This will explain your configuration. When you're happy, click Save.
    

The alert flow

Inside the alert flow you will have access to a number of variables coming from the alerting service:

• flowName - the name of the flow triggering the alert
• relativeFailureThreshold - the relative threshold alert configuration
• absoluteFailureThreshold - the absolute threshold alert configuration
• timeRange - the time range alert configuration
• numberOfDetectedFailures - the exact number of flows that failed within the analyzed interval
• intervalUnderAnalysis - the time interval whose analysis triggered the alert - it will be a from - to interval, in UTC timezone. E.g. from 08 Jun 21 06:28 UTC to 08 Jun 21 06:30 UTC
• linkToFailingFlowLogs - the link to the logs of the flow that triggered the alert. It will automatically filter to only show the failed flows

The default alerting flow will send an email to the specified email address but you can customize it:

• change the notified emails
• change the channel of notification (add WhatsApp or SMS notifications)
• change the message of notification

At runtime, the default flow will generate an email following this template:

The alert conditions for the "{{flowName}}" flow have been met: {{numberOfDetectedFailures}} errors encountered {{intervalUnderAnalysis}}. 
<br/>
Find out more about the failure reasons by looking at the <a href="{{linkToFailingFlowLogs}}">logs</a>.

Which will look something like this: