Retrying HTTP requests

[gojo (qwerty)]

I need to make quite a few http GET/POST requests in my script so I moved the request part to a package and added simple retry in case the first request fails.
This is my code:

func Request(req *http.Request, someclient *http.Client, counter int) ([]byte, error) {
	resp, httperr := client.Do(req)
	if httperr != nil {
		switch counter {
		case 0:
			fmt.Println("Retrying second time...")
			counter += 1
			return Request(req, client, counter)
		case 1:
			fmt.Println("Retrying third time...")
			counter += 1
			return Request(req, client, counter)
		case 2:
			return nil, fmt.Errorf("Couldn't get a response from remote server: ", httperr)
		}
	}
	body, err := ioutil.ReadAll(resp.Body)
	if err != nil {
		return nil, fmt.Errorf("Error while reading the response body: ", err)
	}
	resp.Body.Close()
	return body, httperr
}

Since I’m a noob to Go and programming in general, could anybody review my function?
Any pointers on making this more robust or cleaner? Like calling time.Sleep for 2 seconds between each recursive call?

[gojo (qwerty)]

Anybody?

I mean I understand my code is noob level, but any help regarding pitfalls with this function or tips to improve are really appreciated.

[calmh (Jakob Borg)]

Sure! Here are some off the cuff thoughts;

• this would be clearer as a loop instead of recursion, not requiring the caller to pass an initial zero
• you should probably look closer at the returned error to determine if it should be retried
• if you do retry, sleeping a bit between is nice
• don’t capitalize errors, and rather wrap using something like github.com/pkg/errors that retains the original error for inspection
• if you want retries, maybe also handle errors from readall as this is where you’ll get some of the network errors etc

[gojo (qwerty)]

Thank you.