Fix NGINX Propagation

Redeploy after NGINX Fix

5 minutes

Redeploy After Saving

Saving the file does not update the running gateway. You must apply the ConfigMap and restart the deployment so NGINX loads the new config.

Run the following:

bash
cd ~/workshop/context-propagation
kubectl apply -f deploy/k8s/gateway-config.yaml
kubectl -n cosmic-shop rollout restart deployment/gateway
kubectl -n cosmic-shop rollout status deployment/gateway --timeout=180s

Verify the pod picked up your changes:

kubectl -n cosmic-shop exec deploy/gateway -- grep traceparent /etc/nginx/conf.d/default.conf

proxy_set_header traceparent $http_traceparent;

Validation Checklist

Run these commands after redeploy completes.

1. Confirm gateway pod restarted

bash
kubectl -n cosmic-shop rollout status deployment/gateway
kubectl -n cosmic-shop get pods -l app=gateway

text
deployment "gateway" successfully rolled out

NAME                       READY   STATUS    RESTARTS   AGE
gateway-xxxxxxxxxx-xxxxx   1/1     Running   0          30s

AGE should be recent (pod was recreated after config change).

2. Confirm ConfigMap contains trace headers

kubectl -n cosmic-shop get configmap gateway-nginx-config -o yaml | grep traceparent

  proxy_set_header traceparent $http_traceparent;

You should see traceparent, tracestate, and baggage in the output.

3. Confirm traceparent reaches order-api

Send a request through the edge gateway with a known trace header:

bash
kubectl -n cosmic-shop port-forward svc/gateway 8888:80 &
sleep 2

TRACE_ID="deadbeefdeadbeefdeadbeefdeadbeef"
curl -s -H "traceparent: 00-${TRACE_ID}-deadc0dedeadbeef-01" \
  -X POST http://localhost:8888/api/orders \
  -H "Content-Type: application/json" \
  -d '{"productId":"eyepiece-plossl-set","quantity":1,"customerEmail":"trace-test@cosmic.shop"}' \
  | python3 -m json.tool

kill %1 2>/dev/null

json
{
    "message": "Order created",
    "order": { "orderId": "ORD-...", ... },
    "traceHint": {
        "traceId": "deadbeefdeadbeefdeadbeefdeadbeef",
        ...
    }
}

The traceHint.traceId should match the trace ID you sent in the traceparent header - confirming the gateway forwarded it to order-api.

Troubleshooting

Here’s some of the potential issues you may encounter in this step & suggested remediation steps.

Click here for Troubleshooting Guidance

Potential Issue 1: Traces still disconnected after fix

Confirm you saved deploy/k8s/gateway-config.yaml and ran make fix-nginx (or the manual kubectl apply + restart commands)
Confirm the gateway pod restarted: kubectl -n cosmic-shop get pods -l app=gateway
Verify the ConfigMap: kubectl -n cosmic-shop get configmap gateway-nginx-config -o yaml | grep traceparent
Verify inside the pod: kubectl -n cosmic-shop exec deploy/gateway -- cat /etc/nginx/conf.d/default.conf
Generate new traffic - old traces won’t change retroactively

Potential Issue 2: ConfigMap updated but pod still shows old config

NGINX does not hot-reload ConfigMap volume changes on an existing pod. Always rollout restart deployment/gateway after editing the YAML.

Potential Issue 3: CORS preflight stripping headers

If you add CORS to NGINX, ensure traceparent and tracestate are in Access-Control-Allow-Headers:

nginx
add_header Access-Control-Allow-Headers 'Content-Type, traceparent, tracestate, baggage';

Redeploy after NGINX Fix

Redeploy After Saving #

Validation Checklist #

1. Confirm gateway pod restarted #

2. Confirm ConfigMap contains trace headers #

3. Confirm traceparent reaches order-api #

Troubleshooting #

Potential Issue 1: Traces still disconnected after fix #

Potential Issue 2: ConfigMap updated but pod still shows old config #

Potential Issue 3: CORS preflight stripping headers #