IssueTracker

Problem you have encountered:

The customer has labeled their namespace with “istio-injection=enabled”. When deploying their pods to this namespace, pod’s Readiness Gates remain pending, and the pod starts to show a lot of “Readiness probes fail GET 503” errors. The issue will resolve itself after 10-15 mins.

Additionally, the issue will appear when meeting the following condition:

“istio-injection=enabled” is labeled in ns. Using NAP + spot node.

The issue DOESN'T happen on:

• Istio-injection label in NS + Non-NAP + spot node
• WITHOUT an istio-injection label in NS + Non-NAP + spot / standard node Error message: Readiness probe failed: HTTP probe failed with statuscode: 503

What you expected to happen:

• DPv2 Observability can be enabled with CSM and not have failing health checks for the hubble-peer service.
• Document workaround for DPv2 observability feature support.

Other information:

After investigation with the product team, we found the issue service, hubble-peer. For now, the best approach would be to remove the health check for this service specifically.