Create New Relic Alert based on permformance degradation (COMPARE WITH Query)

Setup Scripted API Monitor

• Create 2 new API keys: an INGEST-LICENSE and USER
• Go to New Relic > Synthetics > Secure credentials and Create secure credential. Create 2 secure credentials, you will need these for later steps
  • YOUR_SECURE_CRED_CONTAINING_INSERT_KEY: your INGEST-LICENSE key
  • YOUR_SECURE_CRED_CONTAINING_QUERY_KEY: your USER key
• Go to New Relic > Synthetics > Create Monitor. Select Endpoint avaiability
• Provide name and select frequency (e.g. 1 hour -> 1 day)
• Select locations (e.g. Sydney), click Write Script
• Enter the content of the modules\Talisker\base_script.js file
• Modify the following:
  • AccountId: your Account ID
  • TASKS[0].id: you can keep it as frontendResponseTime or change it to anything you want (you will need this later)
  • TASKS[0].name: you can name it anything you want
  • TASKS[0].query: this is the NRQL which is used to query the application performance and compare it with previous period => modify it as required
• click on Validate => make sure it is successful
• if successful , click on Save monitor

Create Alerts

• Click on Query your data and run this query: from Metric select latest(talisker.value) where talisker.monitorId='app-error-count' where talisker.id = 'frontendResponseTime'

• It should return % Difference (((current - previous) / current) * 100) from your NRQL above. So if the value is > 0, this means response time in the last 1 hour is slower than 1 day ago. If it is < 0, it is faster

• Click on Create Alert and configure the threshold accordingly, e.g. raise critical violation if the value is > 10% for at least 5 minutes

• Example incidents: