The problem has been initially caused by the reception of a message from Facebook having a format that our application could not process. As a consequence the application kept on restarting on this specific message, preventing new incoming messages to be processed after that.
1) When the issue has been identified, an intervention has been performed in order to prevent the service from crashing. 2) Right after an evolution has been developed and pushed to production to support the message format that caused the problem.
Thanks to these actions this problem should not occur in the future.