Why does Phoenix (ecto/Postgresx) fail to Connect in dev

Question

I am beginning my Elixir/Phoenix journey and having some trouble with my postgres connection.

When I start up my server I get:

 $ mix phoenix.server
 [error] Postgrex.Protocol (#PID<0.214.0>) failed to connect: ** (Postgrex.Error) tcp connect: connection refused - :econnrefused
 [error] Postgrex.Protocol (#PID<0.217.0>) failed to connect: ** (Postgrex.Error) tcp connect: connection refused - :econnrefused
 [error] Postgrex.Protocol (#PID<0.218.0>) failed to connect: ** (Postgrex.Error) tcp connect: connection refused - :econnrefused
 [error] Postgrex.Protocol (#PID<0.211.0>) failed to connect: ** (Postgrex.Error) tcp connect: connection refused - :econnrefused
 [error] Postgrex.Protocol (#PID<0.215.0>) failed to connect: ** (Postgrex.Error) tcp connect: connection refused - :econnrefused
 [error] Postgrex.Protocol (#PID<0.219.0>) failed to connect: ** (Postgrex.Error) tcp connect: connection refused - :econnrefused
 [error] Postgrex.Protocol (#PID<0.216.0>) failed to connect: ** (Postgrex.Error) tcp connect: connection refused - :econnrefused
 [error] Postgrex.Protocol (#PID<0.213.0>) failed to connect: ** (Postgrex.Error) tcp connect: connection refused - :econnrefused
 [error] Postgrex.Protocol (#PID<0.212.0>) failed to connect: ** (Postgrex.Error) tcp connect: connection refused - :econnrefused
 [error] Postgrex.Protocol (#PID<0.210.0>) failed to connect: ** (Postgrex.Error) tcp connect: connection refused - :econnrefused
 [info] Running Rumbl.Endpoint with Cowboy using http://localhost:4000
 [error] Postgrex.Protocol (#PID<0.215.0>) failed to connect: ** (Postgrex.Error) tcp connect: connection refused - :econnrefused

Being new to Elixir, Phoenix, and Ecto I am unclear how to debug this problem. Any suggestions as to what my problem is or how I might go about debugging it would be much appreciated.

My app's set up

I have a basic app

mix phoenix.new rumbl
cd rumbl
mix deps.get
mix deps.compile

My config/dev.exs has the following db setup

# Configure your database
config :rumbl, Rumbl.Repo,
  adapter: Ecto.Adapters.Postgres,
  username: "elixir",
  database: "rumbl_dev",
  hostname: "localhost",
  pool_size: 10

When I run mix ecto.create there are no errors and I can see the rumbl_dev when I run \l in psql. It is owned by the elixir user too.

Running mix ecto.migrate throws the same connections errors

My pg_hba.conf file has the following

local   all             all                                     trust
# IPv4 local connections:
host    all             all             127.0.0.1/32            trust
# IPv6 local connections:
host    all             all             ::1/128                 trust

Here is what I see when I log in with psql

$ psql --dbname=rumbl_dev --username=elixir --host=127.0.0.1 --password
Password for user elixir:
psql (9.4.5)
Type "help" for help.

rumbl_dev=>

Solution 1

Ok, so I figured it out. It is a simple mistake on my part in the end. Though an easy one to make.

I am using Boxen on my Mac and it changes the port to 15432 for some reason.

I may have landed on this sooner if the mix ecto.create had failed. Not sure why that works.

Hopefully this will help others in the future

Solution 2

This happened to me twice, after macOS crashed.

Most recently I used this:

Fix 1

1. Run postgres -D /usr/local/var/postgres

2. Copy the number after PID if you see something like this:

   FATAL:  lock file "postmaster.pid" already exists 
   HINT:   Is another postmaster (PID 379) running in data directory "/usr/local/var/postgres"?
   

3. kill -9 PID with the process # in place of PID.

---

The steps that worked for me previously:

Fix 2

brew update
brew upgrade
brew postgresql-upgrade-database

Solution 3

• Delete the database rumbl_dev in postgresql to make a fresh start.

• If you like, you can try md5 auth for the dev version on localhost by adding the following line to the pg_hba.conf

  # host DATABASE USER ADDRESS METHOD
  
  host rumbl_dev elixir localhost md5
  

  Note: See here for more on setting up the /etc/postsgresql/9.4/pg_hba.conf and change the settings as you see fit.

• Add the complete settings with password to dev.exs:

  config :rumbl, Rumbl.Repo, 
  adapter: Ecto.Adapters.Postgres, 
  username: "elixir", 
  database: "rumbl_dev", 
  hostname: "localhost", 
  password: "***password***",
  pool_size: 10
  

• Try running mix do ecto.create, ecto.migrate and see how it goes.

Hope this helps, if not, I'm out of ideas here.

Comments

[stephen_m]

Can you try adding the password to the config and see if that helps. config :rumbl, Rumbl.Repo, adapter: Ecto.Adapters.Postgres, username: "elixir", database: "rumbl_dev", hostname: "localhost", password: "***password***" pool_size: 10

[Will]

I tried that too but get the same results - thanks for the suggestion @stephen_m

[stephen_m]

can you try running the following command from the terminal: psql --dbname=rumbl_dev --username=elixir --host=127.0.0.1 --password

[Aleksei Matiushkin]

hostname: "localhost" ⇒ hostname: "127.0.0.1" might help.

[Will]

@stephen_m I appended the result of that to my question. It lets me in no problem.

[Will]

@mudasobwa I had tried that but same result

[Aleksei Matiushkin]

After all, do you have {:phoenix_ecto, "~> 3.0"}, {:postgrex, ">= 0.0.0"} in your mix.exs’ deps?

[stephen_m]

is this line somewhere in your config/dev.exs? --- config :rumbl, ecto_repos: [Rumbl.Repo], if not can you add it and try ecto migration then?

[Will]

@stephen_m that was in my config/config.exs file but I tried it in config/dev.exs also but got the same result unfortunately

[Will]

@mudasobwa my mix.exs has exactly those in the lock they come through as "postgrex": {:hex, :postgrex, "0.11.2", ...} and "phoenix_ecto": {:hex, :phoenix_ecto, "3.0.1", ...}

[Will]

same result I am afraid :(

[stephen_m]

glad it worked. don't know why mix ecto.create part works though. you could check it on the console too for the port psql --dbname=rumbl_dev --username=elixir --host=127.0.0.1 --port=5432 --password

[Will]

@stephen_m - it does not work on the console with port=5432. It is odd that the create works.

[neo]

great! thanks so much, I'm using boxen as well. specifying port: 15432 on config.dev.exs solved it.

[Daryl Spitzer]

Fix 1 worked for me (after macOS crashed). Thank you.

[Apie]

Same here. My macOS crashed. Seems to be happening more recently. Came back to postgres giving the same problem and this Fix 1 fixed it for me. The accepted answer seems like a different issue.