-
Expected Behavior Want to be able to authorize request using opa and rego policy defined in a bundle server running in Azure storage account Actual Behavior Steps to Reproduce the Problem
Steps to repro the issue
Authorization Policy apiVersion: security.istio.io/v1beta1
kind: AuthorizationPolicy
metadata:
name: httpbinary
namespace: opa-knative
spec:
selector:
matchLabels:
app: httpbinary
action: CUSTOM
provider:
name: "opa.opa-knative"
rules:
- to:
- operation:
notPaths: ["/ip"] Configmap apiVersion: v1
data:
mesh: |-
defaultConfig:
extensionProviders:
- name: "opa.opa-knative"
envoyExtAuthzGrpc:
service: "opa.opa-knative.svc.cluster.local"
port: "9191" Additional Info Error Logs:
{
"severity": "WARNING",
"timestamp": "2022-11-17T13:51:46.558659653Z",
"logger": "activator",
"caller": "net/revision_backends.go:342",
"message": "Failed probing pods",
"commit": "e82287d",
"knative.dev/controller": "activator",
"knative.dev/pod": "activator-dd544f9cc-rbqkf",
"knative.dev/key": "opa-knative/httpbinary-00001",
"curDests": {
"ready": "10.244.4.128:8012",
"notReady": ""
},
"error": "unexpected status code: want [200], got 403"
} |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 1 reply
-
Found a workaround. The activator probe keeps failing due to the opa policy constantly try to complete a health check as such it needs to confirm that the opa pods are healthy before the activator can send traffic. So disabling the
|
Beta Was this translation helpful? Give feedback.
Found a workaround. The activator probe keeps failing due to the opa policy constantly try to complete a health check as such it needs to confirm that the opa pods are healthy before the activator can send traffic. So disabling the
/healthz
on the opa CUSTOM policy fixed the issue.