What DevOps metrics should you track to measure deployment health?

The four DORA metrics — Deployment Frequency, Lead Time for Changes, Change Failure Rate, and Mean Time to Recovery — are useful from a management point of view, but they do not tell you anything about a deployment and its health as it happens. See What are the DORA metrics? for what those four measure.

For deployment health, watch these instead:

• Healthcheck status per task/node — if this fails, it is reasonable for an automated rollback to trigger.
• Key web page response time(s) — a slowdown here is often the first sign something is wrong.
• All web page response HTTP codes (200, 300, 400, 500) — a shift in the mix of status codes signals trouble before users start complaining.
• Log metrics, such as the number of Warn or Error log lines.

Any deviation in these metrics from their baseline is a sign that the deployment is going sideways.