graphql-engine/server/src-rsr/migrations/42_to_43.sql
Rakesh Emmadi 29f2ddc289 server: support separate metadata database and server code setup for multi sources (#197)
This is an incremental PR towards https://github.com/hasura/graphql-engine/pull/5797

Co-authored-by: Anon Ray <ecthiender@users.noreply.github.com>
GitOrigin-RevId: a6cb8c239b2ff840a0095e78845f682af0e588a9
2020-12-28 12:56:55 +00:00

69 lines
3.1 KiB
SQL

-- This migration enables metadata separation
-- Drops all tables which stores hasura metadata
DROP VIEW hdb_catalog.hdb_role;
DROP TABLE hdb_catalog.hdb_custom_types;
DROP TABLE hdb_catalog.hdb_action_permission;
DROP TABLE hdb_catalog.hdb_action;
DROP VIEW hdb_catalog.hdb_computed_field_function;
DROP TABLE hdb_catalog.hdb_computed_field;
DROP TABLE hdb_catalog.hdb_allowlist;
DROP TABLE hdb_catalog.hdb_query_collection;
DROP VIEW hdb_catalog.hdb_function_info_agg;
DROP VIEW hdb_catalog.hdb_table_info_agg;
DROP TRIGGER hdb_schema_update_event_notifier ON hdb_catalog.hdb_schema_update_event;
DROP FUNCTION hdb_catalog.hdb_schema_update_event_notifier();
DROP TABLE hdb_catalog.hdb_schema_update_event; -- https://github.com/hasura/graphql-engine/pull/6173
DROP TABLE hdb_catalog.remote_schemas;
DROP VIEW hdb_catalog.hdb_function_agg;
DROP TABLE hdb_catalog.hdb_function;
DROP TABLE hdb_catalog.event_triggers;
DROP FUNCTION hdb_catalog.inject_table_defaults(text, text, text, text);
DROP VIEW hdb_catalog.hdb_primary_key;
DROP VIEW hdb_catalog.hdb_unique_constraint;
DROP VIEW hdb_catalog.hdb_check_constraint;
DROP VIEW hdb_catalog.hdb_foreign_key_constraint;
DROP VIEW hdb_catalog.hdb_permission_agg;
DROP TABLE hdb_catalog.hdb_permission;
DROP TABLE hdb_catalog.hdb_remote_relationship;
DROP TABLE hdb_catalog.hdb_relationship;
DROP TRIGGER event_trigger_table_name_update_trigger ON hdb_catalog.hdb_table;
DROP FUNCTION hdb_catalog.event_trigger_table_name_update();
DROP TABLE hdb_catalog.hdb_table;
DROP FUNCTION hdb_catalog.check_violation(text);
-- Remove foreign key constraint to hdb_cron_triggers in hdb_cron_events
ALTER TABLE hdb_catalog.hdb_cron_events DROP CONSTRAINT hdb_cron_events_trigger_name_fkey;
-- Now drop hdb_cron_triggers
DROP VIEW hdb_catalog.hdb_cron_events_stats;
DROP TABLE hdb_catalog.hdb_cron_triggers;
-- Create table which stores metadata JSON blob
CREATE TABLE hdb_catalog.hdb_metadata
(
id INTEGER PRIMARY KEY,
metadata JSON NOT NULL
);
-- DROP hdb_views schema (https://github.com/hasura/graphql-engine/pull/6135)
DROP SCHEMA IF EXISTS hdb_views CASCADE;
-- Note [Migration of schema related to table event triggers log]
-- Table event triggers log related schema is
-- - TABLE hdb_catalog.event_log
-- - TABLE hdb_catalog.event_invocation_logs
-- - PROCEDURE hdb_catalog.insert_event_log
-- We define this schema in any pg source to support table event triggers.
-- There's a possibility of using metadata storage database as a source
-- (more likely if server is started with only --database-url option).
-- In this case, dropping the schema in this up (42 to 43) migration and re-creating the
-- same while defining as a pg source causes loss of event trigger logs.
-- To avoid this we won't drop the schema in this migration. While defining
-- a pg source we will define this schema only if this doesn't exist. This also
-- raises a question, "What happens if old database is only used as metadata storage?".
-- Then, definitely, this schema will be of no use. But, this helps a lot in down
-- migration (opposite to this migration, 43 to 42) as we create this schema only if this
-- doesn't exist.