Istio sidecar 让 pod 重新启动

Question

I've installed on my Openshift cluster the 'Red Hat Openshift Service Mesh' operator, which includes Istio. Now, after adding to my Helm chart the annotation below, a few pods are restarting 2/3 times before properly running.

sidecar.istio.io/inject: 'true'  

Without the annotation, no pods restart take place. In the pod events I see the following errors:

Startup probe failed: HTTP probe failed with statuscode: 503
Readiness probe failed: Get "http://10.253.20.228:15021/healthz/ready": dial tcp 10.253.20.228:15021: connect: connection refused
Startup probe failed: Get "http://10.253.20.228:8080/actuator/health/liveness": dial tcp 10.253.20.228:8080: connect: connection refused

So far I've tried to increase all the possible values of startup/liveness/readiness probes, adding initalDelaySeconds and even the following annotations with no luck:

status.sidecar.istio.io/port : "0"
sidecar.istio.io/rewriteAppHTTPProbers: "false"
sidecarInjectorWebhook.rewriteAppHTTPProbe: "true"

I'm out of ideas, could anybody help?

Answer 1

Maybe try:

proxy.istio.io/config: '{ "holdApplicationUntilProxyStarts": true }'

https://istio.io/latest/docs/reference/config/istio.mesh.v1alpha1/#ProxyConfig-hold_application_until_proxy_starts