Goutham Sekar

We have instrumented the application using newrelic java agent, and also monitoring it using prometheus micrometer metrics. We saw that on both the places the latency were matching and much...

So we use a custom plugin (for auth purposes) and when we removed it we saw improvements, the kong reported upstream latencies were comparable to the actual. But this plugin...

So the plugin intercepts every request made to kong and then makes a HTTP POST request to a custom auth service which has authorization logic. The plugin script itself is...

[middleman-plugin-cm.zip](https://github.com/user-attachments/files/16017833/middleman-plugin-cm.zip) Attached the configMap as a zip Below is the sample plugin yaml ``` apiVersion: configuration.konghq.com/v1 config: response: table timeout: 2 url: http:/// kind: KongPlugin metadata: annotations: kubernetes.io/ingress.class: name: middleman-plugin...

@hanshuebner Thanks for the inputs, will try this setup on our end and get back.

@hanshuebner I setup the test environment, but I set it up on a k8s cluster as that's where we run our tests. First I ran a 15 min test and...

I was having a look at the response headers which gives the X-Kong-Upstream-Latency and X-Kong-Proxy-Latency. The X-Kong-Proxy Latency is actually very consistent, around below 220ms which is expected, but the...

@hanshuebner Update from my end, looks like it is a plugin issue, updated the script to use cjson instead of the json script I had shared and made use of...

Attaching the updated plugin scripts here. You could continue to use the same test env that was previously setup. All I want to check is whether the metrics reported by...