Expose convenient API for verbose logging with args about backoff HOT 5 CLOSED

@k-funk I think I'm going to make regular retries (on_backoff) log as INFO rather than ERROR for 1.5

from backoff.

+1
I've put logging.getLogger('backoff').setLevel(logging.CRITICAL) into my application because I find the default logging to be too verbose.

def giveup_handler(details):
    raise Exception('Too many tries: {}'.format(details['tries']))

@backoff.on_predicate(backoff.expo, predicate_fn, max_tries=3, on_giveup=giveup_handler)
def get_item():
    pass

Somewhat Related:
Since the max_retries kwarg doesn't make the function actually raise an Exception, I have to write that silly def giveup_handler, but then the details still get logged even though I'm handling the giveup myself, which I find odd.

from backoff.

Hm for your second point, when max_tries is exhausted, the exception will be let through. Are you not seeing that?

>>> import backoff
>>> @backoff.on_exception(backoff.expo, Exception, max_tries=3)
... def foo():
...     raise Exception('foo')
...
>>> foo()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File ".../backoff/backoff/_sync.py", line 141, in retry
    ret = target(*args, **kwargs)
  File "<stdin>", line 3, in foo
Exception: foo

from backoff.

Not referring to my fn raising an exception. I'm mostly referring to the backoff decorator: it doesn't raise an exception when max_tries is reached, which seems to be by design. I just find this odd because it logs as if an exception was thrown (logging.error).

from backoff.

Here is another vote for lower priority default logging for exceptions.
#33

from backoff.