distributed sqlite with "old databases"

What is the idea ?

When running ara in distributed sqlite mode against a few databases of mine it complained that a few fields are missing. This makes sense since the databases aren't upgraded at all. I am wondering if it would make sense (or at least support via a flag) to migrate old databases so they can be viewed with a newer ARA server.

This would probably also mean that we shouldn't ever squash and remove old migrations, but I guess that would be doable.

What do you think?

I've run into this issue as well and I think I manually migrated the databases with something like:

export ARA_DATABASE_NAME=/path/to/ansible.sqlite
ara-manage migrate

It is less than ideal, I will agree.

There are some cases where we automatically run SQL migrations, notably:

when the callback runs with the default offline client
when starting an ara container image
when using the ara_api ansible role

In the client, it is enabled by a flag which defaults to true:

ara/ara/clients/utils.py

Line 18 in dc4344e

run_sql_migrations=True,

ara/ara/clients/offline.py

Lines 27 to 29 in dc4344e

    
           if run_sql_migrations: 
        
               # Automatically create the database and run migrations (is there a better way?) 
        
               execute_from_command_line(["django", "migrate"])

We could revisit this approach or think of something in specific for the sqlite middleware backend.

	if run_sql_migrations:
	# Automatically create the database and run migrations (is there a better way?)
	execute_from_command_line(["django", "migrate"])