We need to enable Dynatrace monitoring on our Kong deployment. We use Helm to install Kong. How can this be achieved?
Enabling Dynatrace to monitor Kong in Kubernetes
Uses:
Kong Gateway
TL;DR
How do I enable Dynatrace runtime instrumentation for Kong when Kong is deployed via Helm?
Set DT_NGINX_FORCE_RUNTIME_INSTRUMENTATION: "on" under customEnv in your Helm values file, keeping the value quoted — an unquoted on is parsed as the YAML boolean and renders as the string "true", which silently fails to enable instrumentation.
Runtime instrumentation also adds 10+ seconds to NGINX startup, so raise your liveness and readiness probe timeouts to accommodate the delay.
Solution
Dynatrace has released new steps to monitor Kong.
We need to add the following to our values file for a Helm install (see the customEnv section in Kong’s Helm chart values.yaml — note the exact line numbers shift between chart releases, so search for customEnv rather than relying on a fixed line anchor):
customEnv:
DT_NGINX_FORCE_RUNTIME_INSTRUMENTATION: "on"Note: quote the value as "on". YAML treats a bare, unquoted on (and off/yes/no) as a boolean — Helm then renders it into the container as the literal string "true" rather than the literal string on that Dynatrace’s agent expects (confirmed via helm template: an unquoted on produces value: "true" in the rendered env var, while a quoted "on" correctly produces value: "on"), so leaving it unquoted will silently fail to enable the instrumentation.
Additionally, the runtime instrumentation adds a notable startup delay (Dynatrace’s own docs cite 10 seconds or more) to NGINX. You will need to adjust your timeouts on your liveness and readiness probes to accommodate this, or Kubernetes may kill the container mid-startup.