zipkin-sample-rate of 0 still samples every request

[Stono]

I have zipkin-sample-rate: "0.0" in my configmap, which is setting the opentracing config correctly:

{
  "service_name": "ingress-nginx-external",
  "collector_host": "zipkin.istio-system.svc.cluster.local",
  "collector_port": 9411,
  "sample_rate": 0
}

However I still observe every single request get sampled:

❯ curl -kvL app.platform-testing.testing.k8.atcloud.io 2>/dev/null | grep -i sampled
    "x-b3-sampled": "1",
/Users/karl.stoney/git/autotrader/charts/helm-infra-calico git/master
❯ curl -kvL app.platform-testing.testing.k8.atcloud.io 2>/dev/null | grep -i sampled
    "x-b3-sampled": "1",

[Stono]

Nginx log:

sample.log

[Stono]

@aledbf whilst we're looking at metrics, any ideas about this one?

[aledbf]

@Stono please test quay.io/aledbf/nginx-ingress-controller:0.412
(contains rnburn/zipkin-cpp-opentracing#27)

[Stono]

Fixed! Thanks @aledbf, however I now have another problem.... I can't "force" a sample:

❯ curl -kvL -H "x-b3-sampled: 1" -H "x-b3-flags: 1" app.platform-testing.testing.k8.atcloud.io 2>/dev/null | grep -i sampled
    "x-b3-sampled": "0",

Any ideas?

[aledbf]

Any ideas?

No. Is this supported?
I just opened opentracing-contrib/nginx-opentracing#54

[Stono]

It is, sending x-b3-flags: 1 should force a sample

[aledbf]

Closing. This is not an issue with the ingress controller itself but related to the nginx-opentracing module.