KnexJS Migration With Associated Seed Data

Question

I'm, in the process of learning BookshelfJS/KnexJS (switching from SequelizeJS), and I'm running into an issue with importing data into multiple tables that were created via the migrations feature within KnexJS. There's 4 tables:

1. servers
2. operating_systems
3. applications
4. applications_servers

With the following constraints:

• servers.operating_system_id references operating_systems.id
• applications_servers.server_id references servers.id
• applications_servers.application_id references applications.id

The tables get created just fine when I run knex migrate:latest --env development servers, it's when I import seeded data into the tables I get an error.

Originally, I organized the seed data for the 4 tables into 4 different files within the directory ./seeds/dev, which is just ${table_name}.js:

1. operating_systems.js
2. servers.js
3. applications.js
4. applications_servers.js

After a bit of debugging, I came to the realization that the error is being generated when the seed data within the file applications_servers.js, since when I take that one out, the other 3 run just fine. Then when I remove the 3 seed files and move applications_servers.js to the ./seeds/dev/ directory and execute knex seed:run, the applications_servers table gets populated just fine. However when I try to import the seeded data of all 4 files at once, I receive the following error:

# knex seed:run
Using environment: development
Error: ER_NO_REFERENCED_ROW_2: Cannot add or update a child row: a foreign key constraint fails (`bookshelf_knex_lessons`.`applications_servers`, CONSTRAINT `applications_servers_server_id_foreign` FOREIGN KEY (`server_id`) REFERENCES `servers` (`id`) ON DELETE CASCADE)
at Query.Sequence._packetToError (/Users/me/Documents/scripts/js/node/bookshelf_knex/node_modules/mysql/lib/protocol/sequences/Sequence.js:48:14)
at Query.ErrorPacket (/Users/me/Documents/scripts/js/node/bookshelf_knex/node_modules/mysql/lib/protocol/sequences/Query.js:83:18)

And not a single row gets inserted, into any table.

I thought that maybe it had something to do with the order that they were being imported. (Since they have to be imported in the order that the files were listed above). So thinking that maybe they were referenced in alpha-numeric order, I renamed them to:

1. 1-operating_systems.js
2. 2-servers.js
3. 3-applications.js
4. 4-applications_servers.js

However, nothing changed, no rows were inserted into any table, so just to be sure, I reversed the sequence of the number prefixes on the files, and again, no changes, not a single row was inserted, and same error returned.

Note: The migration script for creating the tables, as well as all of the seed data, was copy and pasted from a JS file I created which creates the same tables and imports the same data using BookshelfJS/KnexJS, but instead of using the migration features, it just does it manually when executed via node. You can view this file here

Any help would be appreciated!

Edit: When I combine all of the seed files within ./seeds/dev into one file, ./seeds/dev/servers.js, everything gets imported just fine. This makes me think that it may be caused by the knex migrations being executed asynchronously, thus the ID's inserted in the pivot table and the servers.operating_system_id may not be inserted yet into the associated tables... If this is the case, is there a way to setup dependencies in the seed files?

Answer 1

Knex.js's seed functionality does not provide any order of execution guarantees. Each seed should be written such that it can be executed in isolation - ie. your single file approach is correct.

If you want to break your individual seed files into submodules, then you might try the following:

// initial-data.js
var operatingSystems = require('./initial-data/operating-systems.js');
var servers = require('./initial-data/servers.js');
exports.seed = function(knex, Promise) {
  return operatingSystems.seed(knex, Promise)
  .then(function () {
    return servers.seed(knex, Promise);
  }).then(function() {
    // next ordered migration...
  });
}

Answer 2

I use the sql-fixtures module to handle FK relation dependencies in my seed file(s).

Imaginary implementation:

const dataSpec = {
  applications_servers: [{
    name: 'My ASP.Net thingie',
    application_id: 'applications:0',
    server_id: 'servers:0'
  }],

  servers: [{
    name: 'My Windows server',
    operating_system_id: 'operating_systems:0'
  }],

  operating_systems: [{
    name: 'Windows Server 2k10'
  }],

  applications: [{
    name: 'My fab web guestbook',
    description: '...'
  }]
}

but it might be superfluous if you declare your dependencies by means of the ORM.

Sometimes however, it is desirable to avoid coupling the DB seeds to your exact current Model implementation.