Skip to main content

Benjamin Oakes

Photo of Ben Oakes

Hi, I'm Ben Oakes and this is my geek blog. Currently, I'm a Ruby/JavaScript Developer at Liaison. Previously, I was a Developer at Continuity and Hedgeye, a Research Assistant in the Early Social Cognition Lab at Yale University and a student at the University of Iowa. I also organize TechCorridor.io, ICRuby, OpenHack Iowa City, and previously organized NewHaven.rb. I have an amazing wife named Danielle Oakes.

Filtering for the PostgreSQL category. Clear

char vs varchar for UUIDs in MySQL and PostgreSQL

by Ben

I spent some time today evaluating whether switching columns that store UUIDs from varchar(255) to char(36) (or binary, etc) would result in any noticeable performance improvement in MySQL. It does seem like it could make an improvement, but not enough to be worth the effort in our case.

What I learned from researching today:

If I’m wrong about what I took away from reading today, please let me know in the comments. I’d love to learn more about this.

PostgresSQL 9.3 (now with more JSON)

by Ben

PostgresSQL: The Other big open-source database has a new release | ZDNet.

It includes some new built-in functionality for handling data serialized as JSON. Sounds really interesting!

That’s not doing quite what you think…

by Ben

I recently helped an intern at Hedgeye work through a problem with a database query. Because I’m working in a separate timezone, I ended up making suggestions through a GitHub pull request. We discussed and decided that what I wrote was self-contained enough that I should re-post so it can help others.

:conditions => ["event_type != ?", 'LOGIN'||'LOGOUT'],

I don’t think this is doing quite what you think…

'LOGIN' || 'LOGOUT' # => 'LOGIN'

So this turns into:

where event_type != 'LOGIN'

I’m guessing you meant to do:

where event_type != 'LOGIN' or event_type != 'LOGOUT'

But, believe it or not, != is a MySQL proprietary extension to SQL. It would probably be best to use something that’s a part of ANSI SQL:

where event_type <> 'LOGIN' or event_type <> 'LOGOUT'
-- alternative:
where event_type not in ('LOGIN', 'LOGOUT')

Because these are literals (not user-provided values), there’s no point in sanitization using ?.

Conclusion:

:conditions => "event_type not in ('LOGIN', 'LOGOUT')",

database configuration does not specify adapter

by Ben

From an answer I wrote for StackOverflow:

Another possible cause:

In Rails 3.2.x, establish_connection has a default argument set from the environment:

From connection_specification.rb:

def self.establish_connection(spec = ENV["DATABASE_URL"])
  resolver = ConnectionSpecification::Resolver.new spec, configurations
  spec = resolver.spec

The way ConnectionSpecification::Resolver works depends on ENV['DATABASE_URL'] giving a nil if not set. (Normally, it would be something like postgres://...).

So, if you happen to have misconfigured DATABASE_URL such that ENV['DATABASE_URL'] == '', that will give you database configuration does not specify adapter.

Why can’t I connect to Heroku Postgres in production?

by Ben

Summary: Switching to Heroku Postgres? If you use a Crane or Kappa database instance, you might run into problems with using a non-standard port when connecting to PostgreSQL in production. (The problem may not happen in your staging environment if you use the dev plan.) Our app is a Padrino-based web service, but the problem described could happen to any framework/code that only expects standard port numbers. Check to make sure the non-standard port is handled correctly.

I hope this helps someone else in a similar predicament! Please let me know in the comments.

Note: Much of the below is based off of a written discussion we had with Heroku support. Thanks for your help, guys! (I’ve changed usernames, passwords, etc., for obvious reasons.)


We recently migrated from a Heroku Shared Database to Heroku Postgres, per their migration instructions.

The steps went as expected for our staging application, but NOT for the production one. This app runs on the Bamboo stack. We could access the PostgreSQL database via heroku pg:psql HEROKU_POSTGRESQL_OUR_COLOR as expected, but when we did heroku console, we got the following error:

>> User.last
PGError: could not connect to server: Connection refused
  Is the server running on host "ec2-12-34-567-890.compute-1.amazonaws.com" and accepting
  TCP/IP connections on port 5432?

  /app/.bundle/gems/ruby/1.8/gems/activerecord-3.1.4/lib/active_record/connection_adapters/postgresql_adapter.rb:1116:in `initialize'
  /app/.bundle/gems/ruby/1.8/gems/activerecord-3.1.4/lib/active_record/connection_adapters/postgresql_adapter.rb:1116:in `connect'
  /app/.bundle/gems/ruby/1.8/gems/activerecord-3.1.4/lib/active_record/connection_adapters/postgresql_adapter.rb:1116:in `connect'
  /app/.bundle/gems/ruby/1.8/gems/activerecord-3.1.4/lib/active_record/connection_adapters/postgresql_adapter.rb:320:in `initialize'

That confused us, as we would expect both psql and console to fail the same way.

We investigated and noticed that HEROKU_POSTGRESQL_OUR_COLOR had a non-standard port (5602):

postgres://username:password@ec2-12-34-567-890.compute-1.amazonaws.com:5602/databasename

To test whether that made a difference, we changed our DATABASE_URL environment variable to have the standard port (5432):

postgres://username:password@ec2-12-34-567-890.compute-1.amazonaws.com:5432/databasename

However, that did NOT make a difference. We also checked config/database.yml, and it seemed to have correct values.

We were left with a production application that could not connect to its database. As a workaround, we promoted the shared database back to being in use:

heroku pg:promote SHARED_DATABASE

We opened a support ticket with Heroku, and later heard back that there were issues with the Heroku-generated config/database.yml. It did not pick up the non-default port number, which is required for Crane. They updated some code on their end to generate the config/database.yml correctly.

We confirmed that the database.yml was created correctly, but unfortunately that wasn’t the entire problem. Our config/database.rb file (used by the Padrino framework) did not handle the non-standard Postgres port either. After realizing that, it was a simple fix to get the database connection working again; we just had to include the port value in the call to ActiveRecord::Base.establish_connection.

(For those keeping score, there was a total of 2 “non-standard port” bugs happening at the same time.)

Out of curiosity, we asked why non-standard port numbers were used. We were told that they’re only used in the Crane and Kappa plans, so one of those would need to be used in staging to have the same setup. The current implementation of Crane and Kappa use a multi-tenant system, so they listen on multiple ports instead of a single port. (Crane and Kappa are the least expensive production database plans.) We were also told the port numbers could change in the future.

Moral of the story: Make sure your app works without depending on the defaults. That’s good advice in general; it would probably be best to test that you don’t depend on tacit default values for anything (databases or otherwise). One way of doing this would be testing with non-standard values when developing.