database.adoc

Compute Platform Database

Table of Contents

• Permissions
• Schemata
  • compute.meta
  • compute.jobs

This library/platform requires a companion PostgreSQL database that will be used for tracking managed jobs.

The schema of the database is configured and maintained by the library itself, thus the library requires credentials for a user that has permission to create, modify, and delete schemata and tables under the target database.

It is recommended that the database not be shared with or used for other purposes.

The database load generated by this library is quite light, so there should be no issue sharing a PostgreSQL instance with other databases.

Permissions

The user the platform uses to connect to the configured database should have effectively all permissions on the target database.

CREATE DATABASE my_database;

CREATE USER my_db_user WITH LOGIN PASSWORD 'somesecretpassword';

GRANT ALL ON DATABASE my_database TO my_db_user;

Schemata

As of version 1 of the platform, all tables managed by the platform will exist under the schema compute.

compute.meta

The compute.meta table holds key/value pairs and is meant to be used solely for storing data relevant to the maintenance of the service, or service internals unrelated to jobs.

Column	Type	Constraints	Purpose
key	VARCHAR(32)	PRIMARY KEY LEN() > 2	Key/value pair key name.
value	VARCHAR(256)	NOT NULL	Key/value pair value.

compute.jobs

Column	Type	Constraints	Purpose
job_id	BYTEA	PRIMARY KEY LEN() == 16	Hash ID of the job.
status	VARCHAR(11)	NOT NULL IN ('queued', 'in-progress', 'complete', 'failed', 'expired')	Current job status.
queue	VARCHAR(32)	NOT NULL	ID/name of the queue the job was submitted to.
config	TEXT (CLOB)		Serialized configuration for the job.
input_files	VARCHAR[]	NOT NULL	Array of the names of files that were included with the job.
created	TIMESTAMP WITH TIME ZONE	NOT NULL	Timestamp for when the job was originally created and queued.
last_accessed	TIMESTAMP WITH TIME ZONE	NOT NULL	Timestamp for when the job was last accessed.
grabbed	TIMESTAMP WITH TIME ZONE		Timestamp for when the job was pulled from the queue to be run.
finished	TIMESTAMP WITH TIME ZONE		Timestamp for when the job was finished (successfully or not)
output_files	VARCHAR[]		Array of the names of files that were created by the job.