Can somone help me understand a good approach to this problem\n```\n    Overview:  \n\t* We have a service like Docusign. \n\t* The service records the id of each fully signed document in a database and then sends out a confirmation notification to the signing user.\n\t* For each successfully sent notificaiton, the service also writes a log in a flat log file on the server.\n\tIssue:\n\t* Due to a bug, 50% of the notifications are missing being sent out.\n    * The missing notifications are not logged in this log file\n    Task:\n    * Given the database that records fully signed documents id and the flat log file that records all the successful notifications, how would you scalably find the missing notifications.\n```\n\nMy approach would be to read the database every hour to find fully signed documents, and then read the flat log files and try diff the results to find the missing notificaitons.\n\nHowever my solution seems very hand-wavy. Can someone please help me answer this better.

Can somone help me understand a good approach to this problem\n\n    Overview:  \n\t* We have a service like Docusign. \n\t* The service records the id of each 