When I use httproute with http add-on, do I need to manually specify backendRefs to keda-add-ons-http-interceptor-proxy?

[ybbiubiubiu]

summary

I installed the HTTP add-on today to test scaling based on request count. I installed the HTTP add-on component using my own values.yaml file, and then wrote httpscaledobject.yaml based on my httpRoute configuration. When I mocked requests, I found that scaling wasn't triggered. After investigation, I discovered that the httproute's backendRefs wasn't pointing to the keda-add-ons-http-interceptor-proxy service. I'd like to ask if this needs to be changed manually or if the operator will automatically detect and change it?
The creation of the httpscaledobject CRD appears normal. However, the keda-add-ons-http-external-scaler log shows the error: "{"error": "there isn't any valid interceptor endpoint"}". I'm wondering if this is the cause.
I've attached my configuration files below.

Veison

keda http-add-on == 0.12.2
keda == 2.16.1

File

httproute:

Kind:         HTTPRoute
Metadata:
  Generation:          1
Spec:
  Hostnames:
    a.b.c.d
  Parent Refs:
    Group:         gateway.networking.k8s.io
    Kind:          Gateway
    Name:          dcloud-gateway
    Namespace:     a
    Section Name:  b
  Rules:
    Backend Refs:
      Group:
      Kind:    Service
      Name:    test-svc
      Port:    80
      Weight:  1
    Filters:
      Type:  URLRewrite
      URL Rewrite:
        Path:
          Replace Prefix Match:  /v1/predict
          Type:                  ReplacePrefixMatch
    Matches:
      Path:
        Type:   PathPrefix
        Value:  /a/b/v1/predict
    Timeouts:
      Request:  3600s

httpscaledobject.yaml

apiVersion: http.keda.sh/v1alpha1
kind: HTTPScaledObject
metadata:
  name: test-http-scaler
  namespace: a
spec:
  hosts:
     -  a.b.c.d
  pathPrefixes:
    - /v1/predict
  
  scaleTargetRef:
    name: test
    kind: Deployment
    apiVersion: apps/v1
    service: test-svc  
    port: 80

  replicas:
    min: 1 
    max: 2  

  scalingMetric:
    requestRate:
      granularity: 1s
      targetValue: 3  
      window: 1m  

[linkvt]

Hi @ybbiubiubiu,

yes you need to manually point your HTTPRoute's backendRefs to the interceptor proxy. The operator won't do this for you right now.

The idea is that all traffic flows through the interceptor, which counts requests for scaling and then forwards them to your actual service:

Gateway → HTTPRoute → interceptor proxy → your app

I found the following "example" in the repo: https://github.com/kedacore/http-add-on/blob/main/examples/xkcd/templates/httproute.yaml

You basically need to update your HTTPRoute's backendRefs to:

backendRefs:
  - kind: Service
    name: keda-add-ons-http-interceptor-proxy
    namespace: keda
    port: 8080

Since your HTTPRoute and the interceptor are in different namespaces, you will also need a ReferenceGrant in the keda namespace:

apiVersion: gateway.networking.k8s.io/v1beta1
kind: ReferenceGrant
metadata:
  name: allow-httproute-from-a
  namespace: keda
spec:
  from:
    - group: gateway.networking.k8s.io
      kind: HTTPRoute
      namespace: a
  to:
    - group: ""
      kind: Service
      name: keda-add-ons-http-interceptor-proxy

Let me know if that helps!

[ybbiubiubiu]

Thank you for your reply. I got it working last Friday. Initially, I couldn't find a corresponding example, so I asked Chatgpt or other models. Some told me I needed to modify it manually, while others said the operator would modify it automatically, so I wasn't sure. Then I checked the Kekedify documentation and it said that http-add-on needs to be modified manually, so I tried it and it worked. I also have two questions: 1. I see that http-add-on is currently in beta and cannot be used in production environments. When will the official version be available? We are eager for this feature to be released. 2. Do you know what Kekedify's pricing structure is? I couldn't find a corresponding pricing information on their official website. Best wishes!

[linkvt]

Hi,

so according to the README the project is in beta stage since 3 years but mostly as there wasn't much activity until the recent months.
Since ~2 months we have new active maintainers in this project (including myself) who are pushing really hard to get this to a production grade level with really good improvements and features (e.g. massive performance boost and reduced memory/CPU usage or header based routing) as also shown by the commit graph:

I can't speak for all maintainers but my goal is to get this project to v1 within the next few months - like around summer time. There are a few more major changes planned like #1501 but afterwards I would probably focus on stability and related topics.

What would really help us is if you have a bit of time to do a quick evaluation, feedback is always very welcome. Maybe you experience a bug, maybe you don't get infos you need from the docs (I added this topic here to #1516 ), maybe you miss a feature - just let us know :)

As I'm not affiliated with kedify and don't know how their HTTP Scaler is priced or can be compared to this project, I guess you would have to check with them directly.

Hope that helps!

[ybbiubiubiu]

Okay, thank you very much. I look forward to the successful production release of http-add-on. thansk again! 杨斌 ***@***.*** 原始邮件 发件人：Vincent Link ***@***.***> 发件时间：2026年3月3日 17:43 收件人：kedacore/http-add-on ***@***.***> 抄送：Unity ***@***.***>, Mention ***@***.***> 主题：Re: [kedacore/http-add-on] When I use httproute with http add-on, doI need to manually specify backendRefs to keda-add-ons-http-interceptor-proxy?(Discussion #1500) Hi, so according to the README the project is in beta stage since 3 years but mostly as there wasn't much activity until the recent months. Since ~2 months we have new active maintainers in this project (including myself) who are pushing really hard to get this to a production grade level with really good improvements and features (e.g. massive performance boost and reduced memory/CPU usage or header based routing) as also shown by the commit graph: image.png (view on web) I can't speak for all maintainers but my goal is to get this project to v1 within the next few months - like around summer time. There are a few more major changes planned like #1501 but afterwards I would probably focus on stability and related topics. What would really help us is if you have a bit of time to do a quick evaluation, feedback is always very welcome. Maybe you experience a bug, maybe you don't get infos you need from the docs (I added this topic here to #1516 ), maybe you miss a feature - just let us know :) As I'm not affiliated with kedify and don't know how their HTTP Scaler is priced or can be compared to this project, I guess you would have to check with them directly. Hope that helps! — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.