context deadline exceeded & bad file descriptor for network calls in worker

[NorseGaud]

Using signal-handler.go, if I put this http call right under the daemon started:

	log.Println("- - - - - - - - - - - - - - -")
	log.Println("daemon started")

	httpClient := &http.Client{
		Timeout: 5 * time.Second,
	}
	req, err := http.NewRequest("GET", "https://google.com", nil)
	if err != nil {
		log.Println("error creating request", "err", err)
		return
	}
	resp, err := httpClient.Do(req)
	if err != nil {
		log.Println("error making GET request to google.com", "err", err)
		return
	}
	defer resp.Body.Close()
	fmt.Println("GET request to https://google.com status:", resp.Status)

I see:

2024/04/07 09:31:20 - - - - - - - - - - - - - - -
2024/04/07 09:31:20 daemon started
2024/04/07 09:31:25 error making GET request to google.com err Get "https://google.com": context deadline exceeded (Client.Timeout exceeded while awaiting headers)

If I put the same into the worker, same problem:

	go func() {
	LOOP:
		for {
			fmt.Println("in worker")
			time.Sleep(time.Second) // this is work to be done by worker.
			select {
			case <-stop:
				break LOOP
			default:
				httpClient := &http.Client{
					Timeout: 5 * time.Second,
				}
				req, err := http.NewRequest("GET", "https://google.com", nil)
				if err != nil {
					log.Println("error creating request", "err", err)
					break
				}
				resp, err := httpClient.Do(req)
				if err != nil {
					log.Println("error making GET request to google.com", "err", err)
					break
				}
				defer resp.Body.Close()
				fmt.Println("GET request to https://google.com status:", resp.Status)
			}
		}
		done <- struct{}{}
	}()

2024/04/07 09:32:43 - - - - - - - - - - - - - - -
2024/04/07 09:32:43 daemon started
in worker
2024/04/07 09:32:49 error making GET request to google.com err Get "https://google.com": context deadline exceeded (Client.Timeout exceeded while awaiting headers)
in worker
2024/04/07 09:32:50 terminating...
2024/04/07 09:32:55 error making GET request to google.com err Get "https://google.com": context deadline exceeded (Client.Timeout exceeded while awaiting headers)
in worker
2024/04/07 09:32:56 daemon terminated

[NorseGaud]

If I use an IP, I see a different error:

2024/04/07 09:37:14 - - - - - - - - - - - - - - -
2024/04/07 09:37:14 daemon started
2024/04/07 09:37:14 error making GET request to google.com err Get "http://1.1.1.1": dial tcp 1.1.1.1:80: connect: bad file descriptor

[NorseGaud]

OK, figured it out. Looks like it's the closeFiles in

func (d *Context) parent() (child *os.Process, err error) {
	if err = d.prepareEnv(); err != nil {
		return
	}

	defer d.closeFiles()

If I comment the defer out, it works.

2024/04/08 10:14:10 daemon started
GET request status: 200 OK
in worker

UPDATE: It worked once and fails any subsequent runs... :(

[NorseGaud]

AHA! It's a premature close/race condition issue! If I add the call in closeFiles, which causes it to be delayed slowly, the childs can spawn properly with the FDs.

func (d *Context) closeFiles() (err error) {
	httpClient := &http.Client{
		Timeout: 5 * time.Second,
	}
	req, err := http.NewRequest("GET", "http://1.1.1.1", nil)
	if err != nil {
		log.Println("error creating request", "err", err)
		return
	}
	resp, err := httpClient.Do(req)
	if err != nil {
		log.Println("error making CLOSE GET request", "err", err)
		return
	}
	defer resp.Body.Close()
	fmt.Println("CLOSE GET request status:", resp.Status)
	cl := func(file **os.File) {
		if *file != nil {
			(*file).Close()
			*file = nil
		}
	}

Adding a sleep for 1 second before closing is a decent solution for now.

[NorseGaud]

The best thing I can find is to wait 2 seconds for the child and worker to fully init before exiting. It's a bit of a change for existing users, but it seems to work and unblock me.