Deployment FAQ

Deployment FAQ

Service fails to start: etcd connection refused

Check that etcd is running and reachable from the service container:

# Test connectivity
nc -zv etcd-host 2379

# Check etcd health
etcdctl --endpoints=http://etcd-host:2379 endpoint health

In Kubernetes, use the etcd service DNS name, not localhost.

bind: address already in use

Another process is listening on the same port. Find and stop it:

lsof -i :8888
kill -9 <PID>

In Kubernetes, ensure no two pods use the same hostPort.

High memory usage after load spike

Check for goroutine leaks:

curl http://localhost:6060/debug/pprof/goroutine?debug=1

Enable pprof in your config:

Log:
  Mode: console
# Add pprof route in main.go:

import _ "net/http/pprof"
go http.ListenAndServe(":6060", nil)

Requests time out under load

1. Check MaxConns — increase if connection pool is exhausted.
2. Check Timeout — too low for your workload.
3. Check downstream RPC timeouts.
4. Check CPU-based load shedding threshold (CpuThreshold).

MaxConns: 50000
Timeout: 5000
CpuThreshold: 950

How to do zero-downtime deployments?

Use Kubernetes rolling deployments with proper readiness probes:

strategy:
  type: RollingUpdate
  rollingUpdate:
    maxSurge: 1
    maxUnavailable: 0
readinessProbe:
  httpGet: {path: /healthz, port: 8888}
  initialDelaySeconds: 5
  periodSeconds: 5