Understanding the Airflow Metadata Database

Overview

Note: This guide has been written and tested with Airflow Airflow 1.10.10. New tables may be introduced in newer versions.

We at Astronomer often get asked about the structure of the underlying Airflow metadata database. The Airflow metadata database stores configurations, such as variables and connections, user information, roles, and policies. It is also the Airflow Scheduler's source of truth for all metadata regarding DAGs, schedule intervals, statistics from each run, and tasks.

Airflow uses SQLAlchemy and Object Relational Mapping (ORM) in Python to connect and interact with the underlying metadata database from the application layer. Thus, any database supported by SQLAlchemy can theoretically be configured to host Airflow's metadata. On Astronomer, each Airflow deployment is equipped with PostgreSQL database for this purpose. The following guide details all of the tables available in the database repository including dependencies and the complete ERD.

Even though we don't recommend modifying the values directly on the database, as doing so might affect dependencies, understanding the underlying structure can be very useful when it comes to building your own reports or querying directly against the database. You can find some useful queries that go directly against these tables in our Useful SQL queries for Apache Airflow guide.

ERD Diagram

The following diagram displays all the tables from the Airflow database.

Airflow DB

Tables

The Airflow metadata database has a total of 30 tables tables are stored on the public schema by default. The following describe the table structure and reference for the Airflow metadata tables.

public.ab_permission

Table Structure:

F-Key	Name	Type	Description
	id	integer	PRIMARY KEY
	name	character varying(100)	UNIQUE NOT NULL

Tables referencing this one via Foreign Key Constraints:

public.ab_permission_view

public.ab_permission_view

Table Structure:

F-Key	Name	Type	Description
	id	integer	PRIMARY KEY
public.ab_permission.id	permission_id	integer	UNIQUE#1
public.ab_view_menu.id	view_menu_id	integer	UNIQUE#1

Tables referencing this one via Foreign Key Constraints:

public.ab_permission_view_role

public.ab_permission_view_role

Table Structure:

F-Key	Name	Type	Description
	id	integer	PRIMARY KEY
public.ab_permission_view.id	permission_view_id	integer	UNIQUE#1
public.ab_role.id	role_id	integer	UNIQUE#1

public.ab_register_user

Table Structure:

Name	Type	Description
id	integer	PRIMARY KEY
first_name	character varying(64)	NOT NULL
last_name	character varying(64)	NOT NULL
username	character varying(64)	UNIQUE NOT NULL
password	character varying(256)
email	character varying(64)	NOT NULL
registration_date	timestamp without time zone
registration_hash	character varying(256)

public.ab_role

Table Structure:

F-Key	Name	Type	Description
	id	integer	PRIMARY KEY
	name	character varying(64)	UNIQUE NOT NULL

Tables referencing this one via Foreign Key Constraints:

public.ab_permission_view_role
public.ab_user_role

public.ab_user

Table Structure:

F-Key	Name	Type	Description
	id	integer	PRIMARY KEY
	first_name	character varying(64)	NOT NULL
	last_name	character varying(64)	NOT NULL
	username	character varying(64)	UNIQUE NOT NULL
	password	character varying(256)
	active	boolean
	email	character varying(64)	UNIQUE NOT NULL
	last_login	timestamp without time zone
	login_count	integer
	fail_login_count	integer
	created_on	timestamp without time zone
	changed_on	timestamp without time zone
public.ab_user.id	created_by_fk	integer
public.ab_user.id	changed_by_fk	integer

Tables referencing this one via Foreign Key Constraints:

public.ab_user
public.ab_user_role

public.ab_user_role

Table Structure:

F-Key	Name	Type	Description
	id	integer	PRIMARY KEY
public.ab_user.id	user_id	integer	UNIQUE#1
public.ab_role.id	role_id	integer	UNIQUE#1

public.ab_view_menu

Table Structure:

F-Key	Name	Type	Description
	id	integer	PRIMARY KEY
	name	character varying(100)	UNIQUE NOT NULL

Tables referencing this one via Foreign Key Constraints:

public.ab_permission_view

public.alembic_version

Table Structure:

F-Key	Name	Type	Description
	version_num	character varying(32)	PRIMARY KEY

public.chart

Table Structure:

F-Key	Name	Type	Description
	id	serial	PRIMARY KEY
	label	character varying(200)
	conn_id	character varying(250)	NOT NULL
public.users.id	user_id	integer
	chart_type	character varying(100)
	sql_layout	character varying(50)
	sql	text
	y_log_scale	boolean
	show_datatable	boolean
	show_sql	boolean
	height	integer
	default_params	character varying(5000)
	x_is_date	boolean
	iteration_no	integer
	last_modified	timestamp with time zone

public.connection

Table Structure:

Name	Type	Description
id	serial	PRIMARY KEY
conn_id	character varying(250)
conn_type	character varying(500)
host	character varying(500)
schema	character varying(500)
login	character varying(500)
password	character varying(500)
port	integer
extra	character varying(5000)
is_encrypted	boolean
is_extra_encrypted	boolean

public.dag

Table Structure:

Name	Type	Description
dag_id	character varying(250)	PRIMARY KEY
is_paused	boolean
is_subdag	boolean
is_active	boolean
last_scheduler_run	timestamp with time zone
last_pickled	timestamp with time zone
last_expired	timestamp with time zone
scheduler_lock	boolean
pickle_id	integer
fileloc	character varying(2000)
owners	character varying(2000)
description	text
default_view	character varying(25)
schedule_interval	text
root_dag_id	character varying(250)

Indexes:

idx_root_dag_id root_dag_id

public.dag_pickle

Table Structure:

Name	Type	Description
id	serial	PRIMARY KEY
pickle	bytea
created_dttm	timestamp with time zone
pickle_hash	bigint

public.dag_run

Table Structure:

Name	Type	Description
id	serial	PRIMARY KEY
dag_id	character varying(250)	UNIQUE#2 UNIQUE#1
execution_date	timestamp with time zone	UNIQUE#2
state	character varying(50)
run_id	character varying(250)	UNIQUE#1
external_trigger	boolean
conf	bytea
end_date	timestamp with time zone
start_date	timestamp with time zone

Indexes:

dag_id_state dag_id, state

public.import_error

Table Structure:

Name	Type	Description
id	serial	PRIMARY KEY
timestamp	timestamp with time zone
filename	character varying(1024)
stacktrace	text

public.job

Table Structure:

Name	Type	Description
id	serial	PRIMARY KEY
dag_id	character varying(250)
state	character varying(20)
job_type	character varying(30)
start_date	timestamp with time zone
end_date	timestamp with time zone
latest_heartbeat	timestamp with time zone
executor_class	character varying(500)
hostname	character varying(500)
unixname	character varying(1000)

Indexes:

idx_job_state_heartbeat state, latest_heartbeat
job_type_heart job_type, latest_heartbeat

public.known_event

Table Structure:

F-Key	Name	Type	Description
	id	serial	PRIMARY KEY
	label	character varying(200)
	start_date	timestamp without time zone
	end_date	timestamp without time zone
public.users.id	user_id	integer
public.known_event_type.id	known_event_type_id	integer
	description	text

public.known_event_type

Table Structure:

F-Key	Name	Type	Description
	id	serial	PRIMARY KEY
	know_event_type	character varying(200)

Tables referencing this one via Foreign Key Constraints:

public.known_event

public.kube_resource_version

Table Structure:

F-Key	Name	Type	Description
	one_row_id	boolean	PRIMARY KEY DEFAULT true
	resource_version	character varying(255)

Constraints:

Name	Constraint
kube_resource_version_one_row_id	CHECK (one_row_id)

public.kube_worker_uuid

Table Structure:

F-Key	Name	Type	Description
	one_row_id	boolean	PRIMARY KEY DEFAULT true
	worker_uuid	character varying(255)

Constraints:

Name	Constraint
kube_worker_one_row_id	CHECK (one_row_id)

public.log

Table Structure:

Name	Type	Description
id	serial	PRIMARY KEY
dttm	timestamp with time zone
dag_id	character varying(250)
task_id	character varying(250)
event	character varying(30)
execution_date	timestamp with time zone
owner	character varying(500)
extra	text

Indexes:

idx_log_dag dag_id

public.serialized_dag

Table Structure:

Name	Type	Description
dag_id	character varying(250)	PRIMARY KEY
fileloc	character varying(2000)	NOT NULL
fileloc_hash	integer	NOT NULL
data	json	NOT NULL
last_updated	timestamp with time zone	NOT NULL

Indexes:

idx_filelo_hash fileloc_hash

public.sla_miss

Table Structure:

Name	Type	Description
task_id	character varying(250)	PRIMARY KEY
dag_id	character varying(250)	PRIMARY KEY
execution_date	timestamp with time zone	PRIMARY KEY
email_sent	boolean
timestamp	timestamp with time zone
description	text
notification_sent	boolean

Indexes:

sm_dag dag_id

public.slot_pool

Table Structure:

Name	Type	Description
id	serial	PRIMARY KEY
pool	character varying(50)	UNIQUE
slots	integer
description	text

public.task_fail

Table Structure:

Name	Type	Description
id	serial	PRIMARY KEY
task_id	character varying(250)	NOT NULL
dag_id	character varying(250)	NOT NULL
execution_date	timestamp with time zone	NOT NULL
start_date	timestamp with time zone
end_date	timestamp with time zone
duration	integer

Indexes:

idx_task_fail_dag_task_date dag_id, task_id, execution_date

public.task_instance

Table Structure:

Name	Type	Description
task_id	character varying(250)	PRIMARY KEY
dag_id	character varying(250)	PRIMARY KEY
execution_date	timestamp with time zone	PRIMARY KEY
start_date	timestamp with time zone
end_date	timestamp with time zone
duration	double precision
state	character varying(20)
try_number	integer
hostname	character varying(1000)
unixname	character varying(1000)
job_id	integer
pool	character varying(50)	NOT NULL
queue	character varying(256)
priority_weight	integer
operator	character varying(1000)
queued_dttm	timestamp with time zone
pid	integer
max_tries	integer	DEFAULT '-1'::integer
executor_config	bytea

Tables referencing this one via Foreign Key Constraints:

public.task_reschedule

Indexes:

ti_dag_date dag_id, execution_date
ti_dag_state dag_id, state
ti_job_id job_id
ti_pool pool, state, priority_weight
ti_state state
ti_state_lkp dag_id, task_id, execution_date, state

Table public.task_reschedule

Table Structure:

F-Key	Name	Type	Description
	id	serial	PRIMARY KEY
public.task_instance.task_id#1	task_id	character varying(250)	NOT NULL
public.task_instance.dag_id#1	dag_id	character varying(250)	NOT NULL
public.task_instance.execution_date#1	execution_date	timestamp with time zone	NOT NULL
	try_number	integer	NOT NULL
	start_date	timestamp with time zone	NOT NULL
	end_date	timestamp with time zone	NOT NULL
	duration	integer	NOT NULL
	reschedule_date	timestamp with time zone	NOT NULL

Indexes:

idx_task_reschedule_dag_task_date dag_id, task_id, execution_date

public.users

Table Structure:

Name	Type	Description
id	serial	PRIMARY KEY
username	character varying(250)	UNIQUE
email	character varying(500)
password	character varying(255)
superuser	boolean

Tables referencing this one via Foreign Key Constraints:

public.chart
public.known_event

public.variable

Table Structure:

Name	Type	Description
id	serial	PRIMARY KEY
key	character varying(250)	UNIQUE
val	text
is_encrypted	boolean

public.xcom

Table Structure:

Name	Type	Description
id	serial	PRIMARY KEY
key	character varying(512)
value	bytea
timestamp	timestamp with time zone	NOT NULL
execution_date	timestamp with time zone	NOT NULL
task_id	character varying(250)	NOT NULL
dag_id	character varying(250)	NOT NULL

Indexes:

idx_xcom_dag_task_date dag_id, task_id, execution_date

Understanding the Airflow Metadata Database

Overview

ERD Diagram

Tables

public.ab_permission

public.ab_permission_view

public.ab_permission_view_role

public.ab_register_user

public.ab_role

public.ab_user

public.ab_user_role

public.ab_view_menu

public.alembic_version

public.chart

public.connection

public.dag

public.dag_pickle

public.dag_run

public.import_error

public.job

public.known_event

public.known_event_type

public.kube_resource_version

public.kube_worker_uuid

public.log

public.serialized_dag

public.sla_miss

public.slot_pool

public.task_fail

public.task_instance

Table public.task_reschedule

public.users

public.variable

public.xcom

Never miss an update from us.

Run production-grade Airflow out-of-the-box with Astronomer.