Updating to Hasura GraphQL Engine v2

Introduction

This page talks about the conceptual changes introduces in Hasura v2 and things to keep in mind while updating from Hasura v1 to v2.

Do reach out to us if you run into any issues while using Hasura v2 or have any questions regarding any changes introduced.

What has changed?

Concepts

The following are the most significant conceptual changes introduced in Hasura v2:

• Hasura v2 can now connect to multiple databases to generate a unified GraphQL API. Each connected database will have a unique "source name" that is used to identify it. Apart from Postgres, connecting to SQL server databases is also now supported.
• Config for connecting a database used to be a startup configuration (i.e. set via env vars/flags) but since v2, it is a part of Hasura Metadata and can be managed dynamically. See connecting databases.
• Hasura Metadata can now be stored in a separate dedicated Postgres DB.

Note

A detailed changelog with all the new features introduced in Hasura v2 is available on the releases page.

Breaking behavior changes

• Semantics of explicit "null" values in "where" filters have changed

  According to the discussion in this GitHub issue, an explicit null value in a comparison input object will be treated as an error rather than resulting in the expression being evaluated to True.

  For example: The mutation delete_users(where: {id: {_eq: $userId}}) { name } will yield an error if $userId is null instead of deleting all users.

  The older behavior can be preserved by setting the HASURA_GRAPHQL_V1_BOOLEAN_NULL_COLLAPSE env var to true.

• Semantics of "null" join values in Remote Schema relationships have changed

  In a Remote Schema relationship query, the Remote Schema will be queried when all of the joining arguments are not null values. When there are null value(s), the Remote Schema won't be queried and the response of the remote relationship field will be null. Earlier, the Remote Schema was queried with the null value arguments and the response depended upon how the Remote Schema handled the null arguments but as per user feedback, this behavior was clearly not expected.

• Order of keys in objects passed as "order_by" operator inputs is not preserved

  The order_by operator accepts an array of objects as input to allow ordering by multiple fields in a given order, i.e. [{field1: sortOrder}, {field2: sortOrder}] but it is also accepts a single object with multiple keys as an input, i.e. {field1: sortOrder, field2: sortOrder}. In earlier versions, Hasura's query parsing logic used to maintain the order of keys in the input object and hence the appropriate order by clauses with the fields in the right order were generated.

  As the GraphQL spec specifies that input object keys are unordered, Hasura v2.0's new and stricter query parsing logic doesn't maintain the order of keys in the input object, removing the previous guarantee that the generated order by clauses specify the fields in the given order.

  For example: The query fetch_users(order_by: {age: desc, name: asc}) {id name age} which is intended to fetch users ordered by their age and then by their name is now not guaranteed to return results first ordered by age and then by their name as the order_by input is passed as an object. To achieve the expected behavior, the query fetch_users(order_by: [{age: desc}, {name: asc}]) {id name age} should be used, which uses an array to define the order of fields to generate the appropriate order by clause.

  This has an effect on query operators that depend on order_by, most notably distinct_on. The distinct_on operator relies on the order of order_by, and therefore is also impacted by this change.

  To avoid confusion and potentially-unwanted behavior, you should use an array value for all order_by clauses.

• Type name for computed fields' input argument has changed

  The name of the computed field input argument has changed from <function_name>_args to <computed_field_name>_<table_name>_args. This change enables adding a root-level tracked function as a computed field which previously would have thrown an input type conflict error.

• Hasura APIs generated by older Hasura versions cannot be added as Remote Schemas to Hasura v2

  With v2.0, some of the auto-generated schema types have been extended. For example, String_comparison_exp has an additional regex input object field. This means if you have a Hasura API with an older Hasura version added as a Remote Schema then it will have a type conflict. You should update all Hasura Remote Schemas to avoid such type conflicts.

• CLI Migrations are executed sequentially, instead of one large transaction

  While applying multiple Migrations, in earlier Hasura CLI versions all migration files were run under one transaction block. i.e. if any migration threw an error, all the previously successfully executed Migrations would be rolled back. With Hasura CLI v2.0, each migration file is run in its own transaction block but all the Migrations are not executed under one. i.e. if any migration throws an error, applying further Migrations will be stopped but the other successfully executed Migrations up till that point will not be rolled back.

• Passing extra GQL params will now throw validation error

  According to the discussion in this GitHub issue, passing unused, required GQL params will now throw a validation error.

  Per the GraphQL spec, any variable - if defined by an operation as required - should be referenced/used in the operation.

  In the example below, the role_type variable is defined but not referenced or used in operation. Thus, executing this query will result in a validation error.

GraphiQL

Query Variables

Request Headers

Documentation Explorer

⚲

No Schema Available

• Deprecation of database specific env vars

  In v2.0, the values of the following env vars are used to define the connection parameters of the default database while updating an existing instance or while starting a fresh instance. During Metadata initialization, their values are moved to the Metadata of the default database as defined here.

  • HASURA_GRAPHQL_PG_CONNECTIONS
  • HASURA_GRAPHQL_PG_TIMEOUT
  • HASURA_GRAPHQL_NO_OF_RETRIES
  • HASURA_GRAPHQL_PG_CONN_LIFETIME
  • HASURA_GRAPHQL_PG_POOL_TIMEOUT
  • HASURA_GRAPHQL_USE_PREPARED_STATEMENTS
  • HASURA_GRAPHQL_TX_ISOLATION
  • HASURA_GRAPHQL_READ_REPLICA_URLS
  • HASURA_GRAPHQL_CONNECTIONS_PER_READ_REPLICA

  Post the initial setup/update once the Metadata is initialized, these env vars can be considered as Deprecated. i.e. Changing or setting values of these env vars will have no impact as the values in the Hasura Metadata are now used to define the connection parameters.

Hasura configuration

• To accommodate changes for storing information for multiple databases, the Hasura Metadata and the Hasura CLI project versions have been bumped from v2 to v3. The v2 versions of the Metadata and CLI project can continue to be used with Hasura v2 instances. Hasura v2 will assume the v2 Metadata and Migrations belong to a database connected with the name default.

• A new mandatory env var HASURA_GRAPHQL_METADATA_DATABASE_URL is now introduced and is mandatory for storing Hasura Metadata.

• Custom env vars can now be used to connect databases dynamically at runtime.

• With support for multiple databases, older database specific env vars have been deprecated. See details

Existing Metadata

HASURA_GRAPHQL_METADATA_DATABASE_URL must be the connection string for where your metadata existed previously.

Moving from Hasura v1 to Hasura v2

Hasura v1 and Hasura v2 compatibility

All existing Metadata and Migrations from a Hasura v1 instance are assumed to belong to a database named default in Hasura v2.

Hence in Hasura v2, a database with name "default" needs to be added to apply Metadata and Migrations from a Hasura v1 instance.

Post adding a database named default, the Hasura v2 instance should behave equivalently to the Hasura v1 instance and all previous workflows will continue working as they were.

Refer to connecting databases to add a database to Hasura v2.

Migrate Hasura v1 instance to Hasura v2

Hasura v2 is backwards compatible with Hasura v1. Hence simply updating the Hasura docker image version number and restarting your Hasura instance should work seamlessly.

Note

In case you happen to have a large number of past cron and Event Trigger logs in your database, this might slow down the update to v2 and might even cause unexpected errors in certain scenarios.

It is highly recommended to clean up past cron and Event Trigger logs data from the database before attempting the update if you have a lot of historical data.

You can take a dump of this data before cleaning up if you wish to keep the log history. This data can be restored back into the DB if required post the update.

Post update steps

Remove deprecated ENV Vars

With support for multiple databases, database connection related config is now stored in Hasura Metadata and the older database specific env vars have been deprecated. See details.

Post update to Hasura v2 these env vars can be removed as they will now be ignored.

This config can now be modified by updating the Hasura Metadata for the connected database(s) via the Console/CLI/APIs.

Upgrade CLI project to enable multiple database and other new features support

Update your Hasura CLI project to config v3 using the steps mentioned in this guide to take full advantages of the features introduced in Hasura v2.

Post upgrading to config v3, the database connection parameters would have been moved to the metadata. Hence it is important to ensure that the same env vars are used for storing database connection strings across all environments and the Metadata being applied also uses the appropriate env vars.

Note

If you do not need multiple database support or any of new features introduced in Hasura v2, like REST endpoints, inherited roles etc., then you can continue to use config v2 project directory and workflows.

Though we would recommend to upgrade to config v3 anyway as it includes some useful directory structure changes.

Updates to CI/CD after updating to Hasura v2

The following commands need to be executed in the specified order to apply Metadata and Migrations in CI/CD workflows

• If using Hasura CLI project in config v2:
  • No changes needed.
  • Run:
    • hasura migrate apply - (apply migrations to the database named "default")
    • hasura metadata apply - (apply metadata to the database named "default")
• If using Hasura CLI project in config v3:
  • Ensure that the same env vars are used for storing database connection strings across all environments and the Metadata being applied also uses the appropriate env vars.
  • Run:
    • hasura metadata apply - (connect Hasura to the databases configured in the metadata)
    • hasura migrate apply --all-databases - (apply the migrations to the connected databases)
    • hasura metadata reload - (make Hasura aware of any newly created database objects in the previous step)

Downgrading back to Hasura v1

In case there are some issues with your Hasura instance post updating to Hasura v2, you can downgrade back to Hasura v1 by reverting the Hasura docker image version and using the downgrade command to revert the Hasura Metadata catalogue changes:

docker run -e HASURA_METADATA_DATABASE_URL=$POSTGRES_URL hasura/graphql-engine:v2.0.0 graphql-engine downgrade --to-v1.3.3

Note

You can downgrade a Hasura v2 instance to Hasura v1 only if there is only one database connected to it.