Stage 3 · Build
Long-Running Daemons & Services
Signal Handling
os/signal, SIGTERM, SIGINT, SIGHUP, context cancellation, and shutdown deadlines.
7 min readGo for SREBuild
Signal Overview
Signals are Unix's inter-process communication mechanism. Go programs handle a subset of signals. Understanding signal handling is essential for building reliable daemons and services.
| Signal | Default Action | Typical Use |
|---|---|---|
| SIGINT | Terminate | Ctrl+C in terminal |
| SIGTERM | Terminate | Graceful shutdown request |
| SIGQUIT | Core dump + terminate | Debugging — dumps goroutine stacks |
| SIGHUP | Terminate | Reload configuration |
| SIGUSR1 | Terminate | Custom: reopen log files |
| SIGUSR2 | Terminate | Custom: binary upgrade |
signal.Notify
Gosignal.notify-usage.go
46 linesLn 1, Col 1Go
signal.Notify sends signals to a channel. Use a buffered channel (size 1) to prevent losing signals. Multiple handlers can listen for different signals. Always handle os.Interrupt (SIGINT) and SIGTERM.
signal.NotifyContext
Gosignal.notifycontext-pattern.go
41 linesLn 1, Col 1Go
signal.NotifyContext creates a context that is canceled when signals are received. This is the cleanest pattern — the entire application uses context cancellation for shutdown. No channels or goroutines needed.
SIGHUP Reload
Gosighup-configuration-reload.go
34 linesLn 1, Col 1Go
SIGHUP traditionally means reload configuration. atomic.Value provides lock-free config access. Store replaces the config atomically. All goroutines see the new config on next access. This pattern supports zero-downtime config reloads.
Signaling Goroutines
Gocascading-shutdown.go
32 linesLn 1, Col 1Go
Context cancellation propagates to all goroutines. errgroup.Wait blocks until all goroutines finish. A shutdown timeout prevents hanging goroutines from blocking the process. This is the standard shutdown pattern.
Shutdown Deadlines
Goshutdown-with-deadline.go
39 linesLn 1, Col 1Go
Shutdown phases: stop accepting connections, drain in-flight requests, stop background workers. Each phase has its own timeout. Kubernetes terminationGracePeriodSeconds must exceed the total shutdown time.
SIGTERM before SIGKILL
Kubernetes sends SIGTERM, waits for terminationGracePeriodSeconds, then SIGKILL. Your application must handle SIGTERM and shut down cleanly within the grace period. If it does not, SIGKILL forces termination.
Mark this lesson complete to store local progress and unlock a cleaner resume path the next time you visit.