"error":"key \"sources\" not found"

[nirajhi]

Version Information

Server Version: hasura/graphql-engine:v2.2.0.cli-migrations-v3 CLI Version (for CLI related issue): v2.2.0

Environment

MAC 12.0.1

Ubuntu 20.04

What is the expected behaviour?

Migrate new changes in the server.

I have been following docs to automate the migration. However the container constantly throws an error:

{
"error":{
                "path":"$.args.metadata",
                "error":"key \"sources\" not found",
                "code":"parse-failed"
            },
}

I have followd the issues in #6813, but the solution doesnot work.

Keywords

What is the current behaviour?

The container gives constant error when trying to start the server.

How to reproduce the issue?

I followed the hasura documentation for the migrations.

1. Create a new table from console
2. Commit all the changes to the github
3. Git pull from another computer.
4. Try docker-compose up -d

version: '3.6'
services:
  postgres:
    image: postgres:13
    restart: always
    volumes:
    - db_data:/var/lib/postgresql/data
    environment:
      POSTGRES_USER: test
      POSTGRES_PASSWORD: password
      POSTGRES_DB: database
  graphql-engine:
    image: hasura/graphql-engine:v2.2.0.cli-migrations-v3
    ports:
      - "8080:8080"
    # extra_hosts:
    #   - "host.docker.internal:host-gateway"
    depends_on:
      - "postgres"
    restart: always
    volumes:
      - ./hasura/migrations:/hasura-migrations
      - ./hasura/metadata:/hasura-metadata
    environment:
      ## postgres database to store Hasura metadata
      HASURA_GRAPHQL_METADATA_DATABASE_URL: postgres://test:password@postgres:5432/database
      ## this env var can be used to add the above postgres database to Hasura as a data source. this can be removed/updated based on your needs
      PG_DATABASE_URL: postgres://test:password@postgres:5432/database
      ## enable the console served by server
      HASURA_GRAPHQL_ENABLE_CONSOLE: "true" # set to "false" to disable console
      ## enable debugging mode. It is recommended to disable this in production
      HASURA_GRAPHQL_DEV_MODE: "true"
      HASURA_GRAPHQL_ENABLED_LOG_TYPES: startup, http-log, webhook-log, websocket-log, query-log
      ## uncomment next line to set an admin secret
      #HASURA_GRAPHQL_ADMIN_SECRET: ######
volumes:
  db_data:

Screenshots or Screencast

Please provide any traces or logs that could help here.

{"type":"http-log","timestamp":"2022-02-16T21:59:17.351+0000","level":"error","detail":{"operation":{"user_vars":{"x-hasura-role":"admin"},"error":{"path":"$.args.metadata","error":"key "sources" not found","code":"parse-failed"},"request_id":"efd55fd4-ae2e-43a8-a847-003b9934fe8f","response_size":84,"request_mode":"error","raw_query":"{"type":"replace_metadata","version":2,"args":{"allow_inconsistent_metadata":true,"metadata":{"version":3}}}\n"},"request_id":"efd55fd4-ae2e-43a8-a847-003b9934fe8f","http_info":{"status":400,"http_version":"HTTP/1.1","url":"/v1/metadata","ip":"127.0.0.1","method":"POST","content_encoding":null}}} time="2022-02-16T21:59:18Z" level=fatal msg="error applying metadata \n{\n "path": "$.args.metadata",\n "error": "key \"sources\" not found",\n "code": "parse-failed"\n}"

[BenoitRanque]

What is the config version of your metadata? You should be able to see this in the version.yaml file

[roryqueue]

@BenoitRanque I'm getting the same error and my config version is 3 if that helps.

[RigoTamas]

Did you manage to solve this? I'm haveing the same issues.

[BenoitRanque]

Typically this issue means your metadata is missing and can't be read from the directory the cli is looking in, could be a bunch of issues. Faced this once and the issue was windows filesystems are not case sensitive but linux filesystems are, and there was a difference in casing somewhere. Again, could be a bunch of stuff from wrong directory to failure mounting the volume. Any chance your metadata directory is empty?

[toddheslin]

Had the same problem but fixed it with hasura metadata export. After running this, hasura metadata apply would work again. This problem occurred for me on a reasonably new project after running hasura migrate apply --down 2, changing some migrations and running hasura migrate apply again.

Edit: I realised that I changed the name of metadata/databases/Snipbot/tables/public_placement.yaml to metadata/databases/Snipbot/tables/public_placements.yaml but didn't update the table reference in metadata/databases/Snipbot/tables/tables.yaml

Hope this helps someone.

[eaf-aim]

Had the same problem but fixed it with hasura metadata export. After running this, hasura metadata apply would work again. This problem occurred for me on a reasonably new project after running hasura migrate apply --down 2, changing some migrations and running hasura migrate apply again.

Edit: I realised that I changed the name of metadata/databases/Snipbot/tables/public_placement.yaml to metadata/databases/Snipbot/tables/public_placements.yaml but didn't update the table reference in metadata/databases/Snipbot/tables/tables.yaml

Hope this helps someone.

This worked for our issue as well after upgrading from 2.0 to 2.7. Thanks!

[clausMeko]

I often have this issue if a file mentioned in tables.yaml or functions.yaml doesn't exist in the file system or is named differently.

Actually a more precise error message would be nice.