A dashboard shows measurements. An alert asks someone to look at a condition that may need a response. A useful alert gives the reader a way to tell what happened and what to check.
Service Monitoring · Lesson 6
Service Alerts
Learn to interpret an alert's condition, time window, and next action.
Reading the Condition
Consider a rule that checks if more than 5 percent of homepage requests in a five-minute window fail. It also requires at least 100 requests so the alert describes a useful sample for this service.
Window 09:10–09:15 UTC
Route /
Requests 120
Failures 12
Twelve failures out of 120 is 10 percent. Both parts of the rule are satisfied. The notice points to the homepage's observed failures.
Would the same rule alert for one failure out of ten requests?
Checking the Time Window
Use the window shown by the alert, even if the current dashboard now looks healthy. The notice may describe an earlier condition. Check whether the signal has returned to its usual range, and whether important user behavior is working again.
The alert arrived at 09:16 and described 09:10 to 09:15. Where should you first look for matching errors?
Taking a Useful Next Step
An alert about failed homepage requests can link to that route's metrics and recent logs. Verify user impact and look for the cause separately. If the service recovered, save enough context to investigate recurring alerts.
Which first action directly checks whether the homepage still fails?
Lesson complete
You finished Service Alerts
You can now read an alert and decide what to verify.
Check the metric, threshold, and minimum sample.
Use the alert's time window to find matching evidence.
Verify the user-facing symptom with an appropriate check.
Keep your learning progress
Create a free account to save this lesson and continue learning on any device.
Create a free account