2 ms·
No, I don't use cpu limits and memory limits are plentiful. Besides, the only difference between working and non-working setups were literally single switch `us
by vbezhenar 3y ago
No, I don't use cpu limits and memory limits are plentiful. Besides, the only difference between working and non-working setups were literally single switch `use-http2` in the ingress controller and it was repeatable experience.
- dilyevsky 3y agoThat is indeed interesting. I’ve used nginx with grpc (which uses http2) years ago and definitely did not see any such latency issues even at 50k qps