Home/ Questions/Q 8721575
In Process

The Archive Base Latest Questions

Editorial Team
  • 0
Asked: 2026-06-13T07:14:38+00:00

My Flask application is currently uses PostgreSQL to store all authentication information (users, tokens)

  • 0

My Flask application is currently uses PostgreSQL to store all authentication information (users, tokens) and interact with it. For other logic I need to use MongoDB. I like to use one technologies of one type instead of multiple to reduce complexity of an application (for example, only Redis instead of Redis + memcached).

So for now I’m thinking about using MongoDB only, using it as backend for authentication process.

Current workflow is following: PostgreSQL stores two tables: User and Token. When I sign up user, I open transaction, store his data (username, login, password) in User table, then insert activation token to Token table, then send activation letter, then close transaction. So, problem occurred in storing User, Token or later in code, when I’ll try to send email, transaction will be rolled back. It prevents cases when user created, but token not, so account can’t be activated.

As I know, transactions is not feature of MongoDB. So, if I will have two documents, User and Token, I will not be able to rollback creating of first if second can’t be created.

My questions are:

  1. How would you implement described behavior on MongoDB?
  2. Is it good idea to use only MongoDB for all stuff instead of PostgreSQL for authentication and MongoDB for documents?
Share

Leave an answer

You must login to add an answer.

Need An Account,

1 Answer

  1. Editorial Team

    Yes, you do have to implement the signup logic yourself in this case. For example if you store the following document:

    user : {
    name : 'Alex Black'
    email : 'alex@example.com'
    token : {
        value : 'some_random_token'     # subject for indexing
        expires : '00.00.00.18.11.2012' # use MongoDB date here
    }
}

    Then e.g. a cron script should remove all the expired users. But a trick would be in using the NoSQL features of Mongo!

    Just create two separate collections: Users and e.g. UnregisteredUsers. Store the user information to UnregisteredUsers and only when registration is confirmed transfer the user document from UnregisteredUsers to Users.

    • 0