Installation on existing cluster fails with "could not initialize jaeger tracer: lookup grafana-agent-traces.default.svc.cluster.local on some-ip:53: no such host"

[mauricewittek]

I tried deploying the app on an existing cluster following the guide. After running the install script, all 3 pods keep crashing. The logs show the same error message for all 3 pods: level=error msg="error initializing tracing" err="could not initialize jaeger tracer: lookup grafana-agent-traces.default.svc.cluster.local on 10.245.0.10:53: no such host"

I tried deploying the app to two different clusters, on on Civo (v1.22.11+k3s1) and the other one on DigitalOcean (Kubernetes 1.23.9-do.0).

[mauricewittek]

My mistake, I had two problems in my cluster:

I didn't check if the grafana-agent-traces Pod was running before deploying the app. The tracer needed 12 minutes to go from containerCreating to running state in my cluster which was unexpected.

After the tracer was running I had to change the value of JAEGER_AGENT_HOST in the 3 deployments as my tracer agent wasn't installed in the default namespace.

Now it works.