Tag: #OpenTelemetry
Kubernetes, DevOps and MLOps, written from things we ran and measured ourselves · 3 posts
gRPC Debugging and Observability: grpcurl, Reflection, channelz, and Metric Pitfalls
gRPC is binary, so you cannot peer into it with curl. Instead, servers act as windows through reflection (which grpcurl uses), standard health services, channelz providing statistics on channels, subchannels, and sockets
2026-10-16 · 18 min read #gRPC#grpcurl#서버 리플렉션#channelz#PrometheusgRPC in Action on a Small Cluster: etcd, CRI, Hubble, OTLP
We measured gRPC by simply reading four conversations on a small cluster with 8 nodes. kube-apiserver had 247 TCP connections open to etcd on the same node because Kubernetes creates a separate etcd client for each resou
2026-10-16 · 19 min read #gRPC#etcd#CRI#containerd#HubbleStreaming, deadline, retry, and keepalive: When calls run long or fail
gRPC methods fall into four categories depending on whether the request and response are single or streamed. Streams are governed by HTTP/2 flow control windows, time limits are propagated to downstream servers as the re
2026-10-16 · 20 min read #gRPC#스트리밍#deadline#재시도#keepalive