Logfire¶

The template traces to Logfire using the same setup as fastapi-environment. It sends nothing until a token is found.

Traces are stored in Logfire. Nothing is written to your databases.

Project¶

environments up prod creates one Logfire project and a write token with Terraform. The project is named <root project>_database. Staging and prod share it. They are split by the deployment.environment attribute.

Terraform needs a Logfire API key with project and token management scopes:

export LOGFIRE_API_KEY=your-api-key

Token¶

Staging and prod read the token from the logfire_token Terraform output. Dev and the migrations database need it set by hand:

export LOGFIRE_TOKEN=your-write-token

In CI, add LOGFIRE_TOKEN as a repository secret. alembic-cicd.yml reads it.

If staging or prod has no token, the CLI prints a warning.

What is traced¶

• Every migrations, models and environments command
• Every step those commands run
• Every alembic run, linked to the command that started it
• Every SQL statement through SQLAlchemy
• Every seed and backfill function
• Python logging records

Settings¶

Custom spans¶

step prints a message and opens a span:

from database_util.utils import step

with step("Loading users...", "load users", count=10):
    ...

logfire is configured before seeds and backfills run. Use it directly:

import logfire

@seed(["dev"])
def users(session: Session) -> None:
    with logfire.span("insert users"):
        ...