Add Discovery Apollo GraphQL server #5145
Conversation
| sourceType: 'module' | ||
| }, | ||
| rules: { | ||
| '@typescript-eslint/explicit-function-return-type': 'off', |
There was a problem hiding this comment.
Will be nice to use shared eslint once monorepo is a thing!
| // } | ||
|
|
||
| // type Track { | ||
| // blocknumber: Int! |
There was a problem hiding this comment.
Oh nice, thanks for creating schemas for all these already!
| @@ -30,6 +29,7 @@ x-common: | |||
| - "audius-protocol-storagev2-1:host-gateway" | |||
| - "audius-protocol-storagev2-2:host-gateway" | |||
| - "audius-protocol-storagev2-3:host-gateway" | |||
| - "audius-protocol-discovyer-provider-graphql:host-gateway" | |||
There was a problem hiding this comment.
| - "audius-protocol-discovyer-provider-graphql:host-gateway" | |
| - "audius-protocol-discovery-provider-graphql:host-gateway" |
| // type TrackSegment { | ||
| // duration: String! | ||
| // multihash: String! | ||
| // } | ||
|
|
||
| // type Followee { | ||
| // user: User! | ||
| // is_delete: Boolean! | ||
| // repost_item_id: String! | ||
| // repost_type: String! | ||
| // } | ||
|
|
||
| // type Download { | ||
| // is_downloadable: Boolean | ||
| // requires_follow: Boolean | ||
| // cid: String | ||
| // } | ||
|
|
||
| // type FieldVisibility { | ||
| // genre: Boolean! | ||
| // mood: Boolean! | ||
| // tags: Boolean! | ||
| // share: Boolean! | ||
| // play_count: Boolean! | ||
| // remixes: Boolean! | ||
| // } |
There was a problem hiding this comment.
nice! did you generate these using a tool or are they hand written?
| const dbConfig = { | ||
| client: 'pg', | ||
| connection: { | ||
| host: 'db', | ||
| port: 5432, | ||
| user: 'postgres', | ||
| password: 'postgres', | ||
| database: 'audius_discovery' | ||
| } | ||
| } |
There was a problem hiding this comment.
it may be simpler to use a connection string like our other knex/ts apps, that way if any of the configuration changes it's only in the url, plus you can gather it right from the discovery .env file
example of how the notifications plugin does it:
https://github.com/AudiusProject/audius-protocol/blob/main/discovery-provider/plugins/notifications/src/conn.ts
| .where('is_current', true) | ||
|
|
||
| if (res.length !== 1) { | ||
| throw new Error('Expected SINGLE row in blocks marked as current') | ||
| } |
There was a problem hiding this comment.
you can add a .first() here and avoid this check as well as the throwing
https://knexjs.org/guide/query-builder.html#first
|
|
||
| export const DiscoveryDB = (dbConnection: Knex) => { | ||
| const blockConfirmationDataLoader = new DataLoader( | ||
| async (blockNumbers: readonly number[]) => { |
There was a problem hiding this comment.
i'm curious, is the readonly annotation required for GQL queries?
| "knex": "2.4.2", | ||
| "lodash": "4.17.21", | ||
| "pg": "8.10.0", | ||
| "sequelize": "6.31.0" |
There was a problem hiding this comment.
are knex and sequelize both required? seems like sequelize could be removed
| @@ -310,6 +311,7 @@ def load_env(protocol_dir, service, environment): | |||
| @click.option( | |||
| "--no-solana", is_flag=True, help="Don't bring up solana-test-validator container" | |||
| ) | |||
| @click.option("--no-graphql", is_flag=True, help="Don't bring up graphql container") | |||
There was a problem hiding this comment.
i know this kind of goes against the established pattern but shouldn't the default here be to not start up the gql container? that way we have to explicitly bring it up until it's ready for release, that also would avoid the need for other to pass this flag just to run their local stack
| ARG git_sha | ||
| ARG audius_loggly_disable | ||
| ARG audius_loggly_token | ||
| ARG audius_loggly_tags | ||
|
|
||
| ENV GIT_SHA=$git_shas | ||
| ENV audius_loggly_disable=$audius_loggly_disable | ||
| ENV audius_loggly_token=$audius_loggly_token |
There was a problem hiding this comment.
do we use loggly anymore?
Description
-Add new container containing an apollo graphql server to be run on discovery nodes
Tests
Monitoring - How will this change be monitored? Are there sufficient logs / alerts?