Open pankajmt opened 3 weeks ago
I am also looking to try max-serving-threads set to something like 30 when cpu requests is set to 3.
We do not set limits in reference to this section in the docs - Gatekeeper uses automaxprocs to default this value to the CPU limit set by the linux cgroup (i.e. the limits passed to the Kubernetes container).
What steps did you take and what happened: [A clear and concise description of what the bug is.]
We are running load test on a policy making an external_data call to a service which is in another region and hence has a 100+ msec latency. validatingWebhookTimeoutSeconds is set to 5. Still, on our load tests, I see validation latencies capped at 3 secs and
denied admission: eval_cancel_error: caller cancelled query execution
errors. Any ideas why?I know if we upgrade to 3.13, we can enable provider caching.
What did you expect to happen:
Cap at 5 sec and then start throwing eval_cancel_error.
Anything else you would like to add: [Miscellaneous information that will assist in solving the issue.]
Environment:
kubectl version
): v1.27.13-eks-3af4770