sentry/bin/split-silo-database

#!/usr/bin/env python
import click
import docker
from django.apps import apps

from sentry.runner import configure
from sentry.silo.base import SiloMode

configure()

from django.conf import settings

from sentry.models.organizationmapping import OrganizationMapping


def exec_run(container, command):
    wrapped_command = f'sh -c "{" ".join(command)}"'
    exit_code, output = container.exec_run(cmd=wrapped_command, stdout=True, stderr=True)
    if exit_code:
        click.echo("Container operation Failed!")
        click.echo(f"Container operation failed with {output}")
    return output


def split_database(tables: list[str], source: str, destination: str, reset: bool, verbose: bool):
    click.echo(f">> Dumping tables from {source} database")
    command = ["pg_dump", "-U", "postgres", "-d", source, "--clean"]
    for table in tables:
        command.extend(["-t", table])
    command.extend([">", f"/tmp/{destination}-tables.sql"])

    client = docker.from_env()
    postgres = client.containers.get("sentry_postgres")

    if verbose:
        click.echo(f">> Running {' '.join(command)}")
    exec_run(postgres, command)

    if reset:
        click.echo(f">> Dropping existing {destination} database")
        exec_run(postgres, ["dropdb", "-U", "postgres", "--if-exists", destination])
        exec_run(postgres, ["createdb", "-U", "postgres", destination])

    citext_command = [
        "psql",
        "-U",
        "postgres",
        destination,
        "-c",
        "'CREATE EXTENSION IF NOT EXISTS citext'",
    ]

    if verbose:
        click.echo(f">> RUNNING: {' '.join(citext_command)}")
    exec_run(postgres, citext_command)

    # Use the dump file to build control silo tables.
    click.echo(f">> Building {destination} database from dump file")
    import_command = ["psql", "-U", "postgres", destination, "<", f"/tmp/{destination}-tables.sql"]
    if verbose:
        click.echo(f">> Running {' '.join(import_command)}")
    exec_run(postgres, import_command)

    if destination == "region" and reset:
        click.echo(">> Cloning stored procedures")
        function_dump = [
            "psql",
            "-U",
            "postgres",
            source,
            "-c",
            "'\\sf sentry_increment_project_counter'",
        ]
        function_sql = exec_run(postgres, function_dump)

        import_function = [
            "psql",
            "-U",
            "postgres",
            destination,
            "-c",
            "'" + function_sql.decode("utf8") + "'",
        ]
        exec_run(postgres, import_function)


def revise_organization_mappings(legacy_region_name: str):
    if settings.SENTRY_MONOLITH_REGION == legacy_region_name:
        click.echo(
            "> No OrganizationMapping have been modified. Set 'SENTRY_MONOLITH_REGION' in sentry.conf.py to update monolith mappings."
        )
    else:
        qs = OrganizationMapping.objects.filter(region_name=legacy_region_name)
        record_count = len(qs)
        qs.update(region_name=settings.SENTRY_MONOLITH_REGION)
        click.echo(
            f"> {record_count} OrganizationMapping record(s) have been updated from '{legacy_region_name}' to '{settings.SENTRY_MONOLITH_REGION}'"
        )


@click.command()
@click.option(
    "--legacy-region-name",
    default="--monolith--",
    help="Previous value of settings.SENTRY_MONOLITH_REGION to overwrite in organization mappings",
)
@click.option("--verbose", default=False, is_flag=True, help="Enable verbose logging")
@click.option(
    "--reset",
    default=False,
    is_flag=True,
    help="Reset the target databases to be empty before loading extracted data and schema.",
)
@click.option("--database", default="sentry", help="Which database to derive splits from")
def main(database: str, reset: bool, verbose: bool, legacy_region_name: str):
    """
    This is a development tool that can convert a monolith database into
    control + region databases by using silo annotations.

    This operation will not modify the original source database.
    """
    # We have a few tables that either need to be in both silos,
    # or only in control. These tables don't have silo annotations
    # as they are inherited from django and their silo assignments
    # need to be manually defined.
    region_tables = ["django_migrations", "django_content_type"]
    control_tables = [
        "django_migrations",
        "django_admin_log",
        "django_content_type",
        "django_site",
        "django_session",
        "auth_user",
        "auth_group",
        "auth_permission",
        "auth_group_permissions",
        "auth_user_groups",
        "auth_user_user_permissions",
    ]
    for model in apps.get_models():
        silo_limit = getattr(model._meta, "silo_limit", None)
        if not silo_limit:
            click.echo(f"> Could not find silo assignment for {model._meta.db_table}")
            continue
        if SiloMode.CONTROL in silo_limit.modes:
            control_tables.append(model._meta.db_table)
        if SiloMode.REGION in silo_limit.modes:
            region_tables.append(model._meta.db_table)

    revise_organization_mappings(legacy_region_name=legacy_region_name)
    split_database(control_tables, database, "control", reset=reset, verbose=verbose)
    split_database(region_tables, database, "region", reset=reset, verbose=verbose)


if __name__ == "__main__":
    main()