task-master/internal/daemon/lifecycle.go

package daemon

import (
	"context"
	"fmt"
	"net"
	"net/http"
	"os"
	"os/exec"
	"path/filepath"
	"syscall"
	"time"

	"task/internal/config"
)

// Status describes whether a daemon is reachable for cfg.DataDir.
type Status struct {
	Running bool
	PID     int
	Socket  string
	DataDir string
}

// StatusOf returns daemon status for the given config.
func StatusOf(cfg *config.Config) (Status, error) {
	p, err := PathsForDataDir(cfg.DataDir)
	if err != nil {
		return Status{}, err
	}
	st := Status{Socket: p.Socket, DataDir: cfg.DataDir}
	pid, err := readPID(p.PIDFile)
	if err == nil && processAlive(pid) {
		st.PID = pid
	}
	if Healthy(p.Socket) {
		st.Running = true
		return st, nil
	}
	return st, nil
}

// Healthy reports whether the daemon HTTP health endpoint responds on socket.
func Healthy(socket string) bool {
	cli := &http.Client{
		Transport: &http.Transport{
			DialContext: func(ctx context.Context, _, _ string) (net.Conn, error) {
				var d net.Dialer
				return d.DialContext(ctx, "unix", socket)
			},
		},
		Timeout: 500 * time.Millisecond,
	}
	resp, err := cli.Get("http://daemon/v1/health")
	if err != nil {
		return false
	}
	defer resp.Body.Close()
	return resp.StatusCode == http.StatusOK
}

// Ensure starts the daemon if needed and waits until it is healthy.
func Ensure(cfg *config.Config) (Paths, error) {
	p, err := PathsForDataDir(cfg.DataDir)
	if err != nil {
		return Paths{}, err
	}
	if Healthy(p.Socket) {
		return p, nil
	}
	cleanupStale(p)
	if err := Spawn(cfg, p); err != nil {
		return Paths{}, err
	}
	deadline := time.Now().Add(5 * time.Second)
	for time.Now().Before(deadline) {
		if Healthy(p.Socket) {
			return p, nil
		}
		time.Sleep(50 * time.Millisecond)
	}
	return Paths{}, fmt.Errorf("daemon did not become healthy at %s", p.Socket)
}

// Stop terminates the daemon for cfg.DataDir if running.
func Stop(cfg *config.Config) error {
	p, err := PathsForDataDir(cfg.DataDir)
	if err != nil {
		return err
	}
	pid, err := readPID(p.PIDFile)
	if err != nil {
		if os.IsNotExist(err) {
			cleanupStale(p)
			return nil
		}
		return err
	}
	if processAlive(pid) {
		if err := syscall.Kill(pid, syscall.SIGTERM); err != nil {
			return fmt.Errorf("signal daemon: %w", err)
		}
		deadline := time.Now().Add(3 * time.Second)
		for time.Now().Before(deadline) {
			if !processAlive(pid) {
				break
			}
			time.Sleep(50 * time.Millisecond)
		}
		if processAlive(pid) {
			_ = syscall.Kill(pid, syscall.SIGKILL)
		}
	}
	cleanupStale(p)
	return nil
}

// Spawn starts a detached daemon process for cfg.
func Spawn(cfg *config.Config, p Paths) error {
	exe, err := os.Executable()
	if err != nil {
		return fmt.Errorf("executable: %w", err)
	}
	logFile, err := os.OpenFile(p.LogFile, os.O_CREATE|os.O_WRONLY|os.O_APPEND, 0o600)
	if err != nil {
		return fmt.Errorf("open log: %w", err)
	}
	defer logFile.Close()

	cmd := exec.Command(exe, "daemon")
	cmd.Env = os.Environ()
	cmd.Stdout = logFile
	cmd.Stderr = logFile
	cmd.Stdin = nil
	cmd.SysProcAttr = &syscall.SysProcAttr{Setsid: true}
	if err := cmd.Start(); err != nil {
		return fmt.Errorf("spawn daemon: %w", err)
	}
	// Detach: do not wait; child writes its own PID file when listening.
	_ = cmd.Process.Release()
	return nil
}

func cleanupStale(p Paths) {
	_ = os.Remove(p.Socket)
	_ = os.Remove(p.PIDFile)
}

// AbsoluteDataDir ensures DataDir is absolute for path hashing.
func AbsoluteDataDir(dataDir string) (string, error) {
	if filepath.IsAbs(dataDir) {
		return filepath.Clean(dataDir), nil
	}
	abs, err := filepath.Abs(dataDir)
	if err != nil {
		return "", err
	}
	return abs, nil
}