Following up on the preliminary investigation on T401131: [SPIKE] Explore monitoring and alarm options for Wikimedia APIs and the work for:
- Implementing better observable data T402074: Set up new prometheus label structure for the API monitoring
- Connecting Slack to alertmanager T401730: Add a pathway for Alertmanager to send alerts in Slack
We now want to add basic alerts for the REST APIs. The planned alert conditions are:
Fire an alert if, out of all requests per minute:
- 500 errors exceed 1%
- 40x errors exceed 20%
- 41x errors exceed 10%
- 42x errors exceed 10%