fix: reset exponential backoff after task runs successfully in tasks …

…extension (Pycord-Development#2700)

* [tasks] reset exponential backoff after task runs successfully

The current behaviour of the _loop method (when the reconnect
attribute is set to True) is to have the same ExponentialBackoff
instance for all of its lifetime. A task that has failed n times
over its life-time will thus wait 2**n seconds before retrying, no
matter how many successful instances of happened between the failures.

With this commit, the behaviour is to reset the exponential backoff
after a succesful execution of the task, so that the backoff only
takes place after successive failures.

* Update CHANGELOG.md

Signed-off-by: DaBlumer <[email protected]>

* Update CHANGELOG.md

Co-authored-by: JustaSqu1d <[email protected]>
Signed-off-by: DaBlumer <[email protected]>

---------

Signed-off-by: DaBlumer <[email protected]>
Co-authored-by: JustaSqu1d <[email protected]>

CHANGELOG.md

@@ -81,6 +81,8 @@ These changes are available on the `master` branch, but have not yet been releas
 - Fixed attachment metadata being set incorrectly in interaction responses causing the
   metadata to be ignored by Discord.
   ([#2679](https://github.com/Pycord-Development/pycord/pull/2679))
+- Fixed unexpected backoff behavior in the handling of task failures
+  ([#2700](https://github.com/Pycord-Development/pycord/pull/2700)).
 
 ### Changed
 

discord/ext/tasks/__init__.py

@@ -168,6 +168,7 @@ async def _loop(self, *args: Any, **kwargs: Any) -> None:
                 try:
                     await self.coro(*args, **kwargs)
                     self._last_iteration_failed = False
+                    backoff = ExponentialBackoff()
                 except self._valid_exception:
                     self._last_iteration_failed = True
                     if not self.reconnect: