Service Health¶
When Azure itself has a problem, you want to hear it from Azure before you spend an hour debugging your own code. Service Health reports Azure incidents, planned maintenance, security advisories and service retirements for the services and regions you use. Resource Health reports the health of individual resources.
Status: planned
ServiceHealth and ResourceHealth events already reach log-platform-prod through the Activity Log export. No Service Health alerts are configured.
What to alert on¶
| Event type | Why | Route to |
|---|---|---|
Service issue (an active incident) in westeurope or global services |
Explains failures that aren't your fault, and tells you when to wait rather than fix | ag-platform-critical |
| Planned maintenance | Plan around it | ag-platform-warning |
| Health advisories | Changes that need action, such as deprecated features | ag-platform-warning |
| Security advisories | Azure-side security issues that may affect you | ag-security |
| Retirements | Services or features being removed. They usually give months of notice, which is easy to miss. | ag-platform-warning, reviewed monthly |
How to deploy it¶
Service Health alerts are Activity Log alerts, and they are per subscription. Two options:
- Policy: the built-in Configure subscriptions to enable service health alert monitoring rule policy creates an alert rule and action group in every subscription. Assign it through EPAC at
mg-intermediate-root, so new subscriptions are covered too (once enforced). This is also part of Azure Monitor Baseline Alerts. - Subscription vending: add the alert rule to the subscription template, so each subscription gets it at creation.
Use one approach, not both.
Resource Health¶
Resource Health tells you whether a specific resource is available, and whether Azure or you caused a problem. Use it:
- during an incident, to rule Azure in or out quickly;
- in alerts for critical resources: a Resource Health alert fires when a resource becomes unavailable or degraded.
Retirement review¶
Once a month, check Service Health → Health advisories and Azure Advisor retirement recommendations. Add anything affecting the platform or the Terraform modules to the backlog, with the retirement date as the deadline.