Lior Franko

This is in the values.yaml of the scrape config, usually Prometheus.

+1

Yes, it happens every time. Here is the SE: ``` apiVersion: networking.istio.io/v1beta1 kind: ServiceEntry metadata: name: sleep-ilo-2-80 namespace: devops-apps spec: endpoints: - address: 10.120.17.59 weight: 1 - address: 10.120.17.180 weight:...

Without SE: ``` { "destinationRules": {}, "name": "sleep-ilo-2", "namespace": "devops-apps", "outboundTrafficPolicy": { "mode": "ALLOW_ANY" }, "rootNamespace": "istio-system", "services": [ ], "sidecar": { "workloadSelector": { "labels": { "app": "sleep-ilo-2" } },...

Sure ``` { "totalClients": 7, "clients": [ { "connectionId": "abir-test-659456cddf-wt6kn.devops-apps-409", "connectedAt": "2024-05-11T14:14:49.560212062Z", "address": "10.120.12.25:41934", "labels": { "app": "abir-test", "clusterName": "general-dev-02", "commit": "1.16.1", "environment": "dev", "kubernetes.io/hostname": "ip-10-120-15-151.ec2.internal", "kyverno-protected": "false", "name": "abir-test",...

Hi Is this currently under development? We have large services with slow pods and fast pods, the slower pods are deployed on slow or busy instances, and the fast pods...

We use Envoy as a sidecar running on Kubernetes with Istio service-mesh. We have services with ~500 pods sending requests to other services with ~500 pods. The environment is a...

> How exactly are you measuring this? Is it expected that you have no in-flight requests? Yes, we're running low-latency services (Up to 30ms response time) Here is a graph...

Thanks for the detailed explanation! All the above makes perfect sense, and when Envoy balances according to the current `upstream_rq_active`, I agree that it's perfectly balanced. Here is the P50...

I think that success latency + error rate can provide a good measurement for weight.