A bit of a weird bug I am experiencing with istio (1.7) on a gke (1.17) cluster.
If I allow istio to rewrite my HTTP probes, I get loads of connectivity errors (and my cluster connectivity is generally really slow):
Readiness probe failed: Get http://10.16.4.64:15021/healthz/ready: dial tcp 10.16.4.64:15021: connect: connection refused Readiness probe failed: HTTP probe failed with statuscode: 503 Readiness probe failed: Get http://10.16.2.106:15021/healthz/ready: net/http: request canceled while waiting for connection (Client.Timeout exceeded while awaiting headers)
This to me sounded like a resource issue, so I gave the cluster more than it needs (almost double) and same events pop up.
Removing the HTTP rewrite does silence most of these problems with only a very few (coming from the healthcheck defined on my containers).
Readiness probe failed: HTTP probe failed with statuscode: 503
remaining, but on the plus side my general connectivity comes back to normal.
Does any of this sound familiar to anyone?
Could anyone point me towards a way to debug this?
PS: these problems were not present before the istio installation