javaagent graceful shutdown

[shintasmith]

We are running our Java app using the OpenTelemetry javaagent, i.e: using -javaagent:$APP_HOME/opentelemetry-javaagent.jar option, inside a Kubernetes pod that also has istio proxy container running.

On pod termination, the javaagent always logs errors (see attached stacktrace)
otel-stacktrace.txt

We tried setting some configuration (terminationDrainDuration) on the istio proxy but that doesn't solve this problem.

How do we configure OpenTelemetry javaagent to be aware of shutdown and to just not flush metrics out?

[jaydeluca]

so from your stack trace the error is

[otel.javaagent 2025-08-07 09:39:52:117 -0500] [OkHttp http://172.27.239.227:4317/...] ERROR io.opentelemetry.exporter.internal.grpc.GrpcExporter - Failed to export metrics. The request could not be executed. Error message: Failed to connect to /172.27.239.227:4317

Which looks like there's some race condition where the connectivity to your collector is broken before the JVM shuts down. When you mention istio, is that because you are proxying your calls to your collector through istio?

[shintasmith]

Yes, in this case the collector runs on the host where the pod runs. istio is proxying all requests going into pod and out of pod.