multi-db | Allows you to partition your large WordPress Multisite | Database library

Django 3.2.6

I'd like some models not to made migrations at all.

Is it possible?

I tried:

1. https://docs.djangoproject.com/en/3.2/ref/models/options/#managed

class Meta: managed = False

2.

I've read the documentation on database routers in Django 2.2. I more or less understand the concept - including keeping certain tables together - except that it seems somewhat tricky to do in practice.

My instincts when things are complicated and interdependent is to use unit tests to gradually get my code to return expected results. Except that I don't know how to write tests in this case.

I've the databases db1 and db2. The schemas of both DBs should be created with the same migration script.

The Django docs mention DATABASE_ROUTERS and RunPython, but I didn't manage to get it working so far. The function is called, but migrations.CreateModel() has no impact when called from the function: the table is not created.

Migration script:

I'm testing out a custom database router for a Django project to support read replicas (based on the Django docs on multiple database) and when I create model instances containing references to other models, the save method tries to use the read-replica, for some reason.

I've registered the following router in the DATABASE_ROUTERS settings:

I have 'default' and 'secondary' databases. Why, even though it is stated:

I have one database with multitples schemas, and I try to use routers to manage record informations. (https://www.amvtek.com/blog/posts/2014/Jun/13/accessing-multiple-postgres-schemas-from-django/)

In my django settings:

I'm working on a django project where I'm using multiple databases with a custom router. Therefore, django should know which database to use for which app.

Now I came across some in my opinion weird behaviour and wanted to ask if anyone knew if that is the correct behaviour or not, especially because the github page doesn't seem to have an issues tracker.

To make the point clear, I'm using an extreme example, which I nonetheless tried. For testing purposes, I set the allow_migrate function to just return False.
I did create a custom router which overrides the allow_migrate function and I also registered it under the DATABASE_ROUTERS settings. The allow_migrate function is also called, as I checked it with some print statements.

The way I understand the documentation for using multiple databases with a custom router, the allow_migration method specifies if a migration is to be applied for the current db, app_label and model_name. If it returns True, the migration is applied, if False, it is 'silently skipped' and therefore not applied.

As I set the allow_migrate function to return False all the time, I expected that none of the yet unapplied migrations are applied.

The migrations are, as the documentation states, indeed 'silently skipped', but the table django_migrations is nonetheless being filled with the not yet applied migrations.

Therefore, if I had an actual logic in the allow_migrate function, even if the migration is correctly not applied, e.g. because the database is being tested for a specific app that should not be stored in that database, the django_migrations table of that database is being filled with wrong data.

So, is that the wanted behaviour? Am I just using the function allow_migrate wrong and should actually be using something else?

In case it is relevant, I'm using django version 3.1.2 and my database is a mariadb database.

I have the following multi-db setup in my rails 6.0.2.2 app:

I am trying to use multiple database where one DB is a default and the second DB is a :memory DB. From the description in the Docs (shown below), it sounds like the default routing should be fine. But it says that the objects are "sticky" to their original database. How is there an "Original" database?

When I first started to implement this, I expected that there would be a META in the Model classes that would specify its database, but that does not seem to be the case. I see examples where people have a Mapping of Apps to DB, which would be perfect for my scenario, but then they turn around and write a Routing anyway.

I don't want to have to always add the database to Save calls, as this would be prone to programming errors. Is there an official setting to map an App to a Database, or a Model to a Database? Or is it always required to write a Router to use multiple databases.