Adding support for disabling default metrics using Tyk Chart

Dear Team,

I have been trying to configure the Prometheus pump using the Tyk helm chart by setting up the environment variables as instructed in the documentation available at Tyk Pump Environment Variables.

However, I have noticed that the TYK_PMP_PUMPS_PROMETHEUS_META_DISABLEDMETRICS environment variable is not functioning as expected. We intended to disable some of the default metrics using this environment variable.

Upon examining the deployment pump chart available at https://github.com/TykTechnologies/tyk-charts/blob/main/tyk-pump/templates/deployment-pmp.yaml, I realized that there is no support for this environment variable. Hence, I would greatly appreciate your help in adding support for this feature.

Thank you for your assistance in this matter.

Hi Kaushal,

Welcome to Tyk Community.
TYK_PMP_PUMPS_PROMETHEUS_META_DISABLEDMETRICS is the correct environment variable. You can define it your values.yaml under pump section as extraEnvs.

Regards,
Asha

1 Like

Hi Asha,

I have tried that but it doesn’t seem to be working. It is still creating default metrics.

e.g.:
- name: TYK_PMP_PUMPS_PROMETHEUS_META_DISABLEDMETRICS
value: ‘[“tyk_latency_bucket”]’

Note that I am using 1.7 version for tyk pump.

Sincerely,
Kaushal

Hi Kaushal,

Currently only base metrics can be excluded. Pls check.
[Tyk Pump Environment Variables]

Regards,
Asha

Hi Asha,

Sorry for the delayed response. I did try disabling the base metrics but the metrics are still exposed.

Below is my configuration

    - name: TYK_PMP_PUMPS_PROMETHEUS_META_DISABLEDMETRICS
      value: '["tyk_latency", "tyk_http_status", "tyk_http_status_per_path", "tyk_http_status_per_key", "tyk_http_status_per_oauth_client"]'

Sincerely,
Kaushal

Can you try inputting the value as a comma separated value instead. For example

tyk_latency, tyk_http_status, tyk_http_status_per_path, tyk_http_status_per_key, tyk_http_status_per_oauth_client

Let us know how it goes

No luck, it is still exposing the metrics.

I’ll check this. I see you mentioned this happens on helm chart in Kubernetes.

I missed this earlier.

@Kaushal_Prajapati Confirmed the behaviour and opened and internal ticket. Will update this thread once it’s fixed.

In the meantime, can you use the config file to enable it

"disabled_metrics": [
          "tyk_http_status",
          "tyk_http_status_per_path",
          "tyk_http_status_per_key",
          "tyk_http_status_per_oauth_client",
          "tyk_latency"
        ],

I tried it and it worked

1 Like

Actually can you double check using comma-separated values? I can see the "Processing metric:tyk_http_status" message stripped in pump logs when debug mode is enabled.

value: >-
  tyk_http_status,tyk_http_status_per_path

Kindly let us know.

It works with config. Thank you!!