fix: graceful shutdown bypasses deferred cleanup due to os.Exit(0)

[mayankpande88]

Problem

In main.go:173, the signal handler calls os.Exit(0) which bypasses all deferred cleanup functions:

• registry.Close() — never called
• tracer.Close() — eBPF programs, uprobes, perf buffers, ring buffers leak
• profiling.Stop() — never called

On every pod restart/rollout, kernel resources (eBPF maps, programs, uprobes) are leaked. On a daemonset that restarts frequently, this compounds.

Fix

Replace os.Exit(0) with context cancellation. Propagate context.Context through the main function and let deferred cleanup run naturally.

ctx, cancel := context.WithCancel(context.Background())
go func() {
    sigChannel := make(chan os.Signal, 1)
    signal.Notify(sigChannel, os.Interrupt, syscall.SIGTERM)
    <-sigChannel
    klog.Infoln("Received signal, shutting down")
    cancel()
}()
// ... use ctx in main loop ...
// deferred Close() calls will run on natural exit

Impact

High — kernel resource leaks on every restart.