[Question] Handling email backend exceptions

[1oglop1]

Hi, thank you for this project it saved me a lot of time!
I have a question about handling exceptions from Anymail and generally any e-mail backend.

My function sends an e-mail:

from django.core.mail import EmailMessage

def dummy():
     email = EmailMessage(
        "subject",
        "body",
        "from@example.com"
        ["to@example.com"],
    )
    email.send()

Is there any general way to catch the exception from email backends other than using broad Exception?
By default Django raises smtplib.SMTPError and anymail has AnymailAPIError.

This would mean that for each email backend I'd have to add its exception in the test:

try:
  email.send()
except SMTPException,AnymailAPIError :
   pass # do something useful

I noticed there was once a proposal for this https://code.djangoproject.com/ticket/23944
I'm wondering if anyone is aware of any development in the Django ecosystem since I'm pretty new to Django.
Even though it's highly unlikely I'll be changing the email backend, I'm still interested in the answer.

Thank you.

[medmunds]

The answer kind of depends on why you're catching the error and what you want to do with it—the specifics of the "do something useful" in your example.

If you're trying to do some generic handling—say, logging the error for later investigation—then catching the broad Exception seems fine. (It's what Django core contributor Tim Graham suggested in closing the Django ticket you found.)

For pretty much anything else, "do something useful" probably involves digging into the details of the error. And that's going to be specific to the email backend and even the particular ESP. For example, you might want to handle a temporary ESP outage by quietly queuing the email to re-send later. But if the error is your API keys are invalid, you'd probably want to handle that with a loud, urgent operational alert. Distinguishing those cases requires examining attributes of the error itself, which will differ between backends and ESPs.

try:
    email.send()
except AnymailError as err:
    # (not all AnymailErrors include an API HTTP status_code)
    status_code = getattr(err, 'status_code', None)
    if 500 <= status_code < 600 or status_code == 429:
        # network error or API rate limit
        queue_email_for_later_retry(email)
    else:
        # some other problem; let logging report the error to us
        raise
except smtplib.SMTPException as err:
    if nontrivial_logic_to_recognize_retryable_smtp_error(err):
        queue_email_for_later_retry(email)
    else:
         raise
# ... etc. ...

I'd be very open to proposals to normalize some of the ESP-specific sending error logic in Anymail. In particular, introducing a way to distinguish transient sending problems (where re-sending the same message later may succeed) to more permanent errors like invalid addresses or sending options.

(Incidentally, the django-mailer project ran into a similar issue. After a multi-year discussion, I believe they came to the conclusion that there isn't a good generic error handling solution across multiple email backends, and ended up adding a user-overridable error handler instead.)

[1oglop1]

Thank you for this elaborate asnwer it helped me a lot!
We are at the stage of building a new service with a lot of open decisions so for now I find the best decision just to let it blow up with an exception and error 5XX. And handle this case later on!