Builds: fix retry (#9133)

We can't rely on `autoretry_for` for exceptions that are raised inside the `before_start` handler.

How autoretry_for works is that it wraps the run method to catch the exceptionhttps://github.com/celery/celery/blob/ab2bcc096a9013a9147a3be1a2699d2312f93d1f/celery/app/autoretry.py#L29-L34But celery calls before_start separately from runhttps://github.com/celery/celery/blob/ab2bcc096a9013a9147a3be1a2699d2312f93d1f/celery/app/trace.py#L445-L448

Closes #9014

Author: stsewd

readthedocs/projects/tasks/builds.py

@@ -296,12 +296,15 @@ def _check_concurrency_limit(self):
             max_concurrent_builds = settings.RTD_MAX_CONCURRENT_BUILDS
 
         if concurrency_limit_reached:
-            # By raising this exception and using ``autoretry_for``, Celery
-            # will handle this automatically calling ``on_retry``
+            # By calling ``retry`` Celery will raise an exception and call ``on_retry``.
+            # NOTE: autoretry_for doesn't work with exceptions raised from before_start,
+            # it only works if they are raised from the run/execute method.
             log.info("Concurrency limit reached, retrying task.")
-            raise BuildMaxConcurrencyError(
-                BuildMaxConcurrencyError.message.format(
-                    limit=max_concurrent_builds,
+            self.retry(
+                exc=BuildMaxConcurrencyError(
+                    BuildMaxConcurrencyError.message.format(
+                        limit=max_concurrent_builds,
+                    )
                 )
             )