Phoenix Ecto, don't connect on kubernetes

Hello all!

Actually I’m getting this error on my kubernetes log:

2017-06-02T02:02:49.13752428Z 02:02:49.137 [error] Postgrex.Protocol (#PID<0.1578.0>) failed to connect: ** (DBConnection.ConnectionError) tcp connect (***.***.***.***:5432): timeout

I already check the secrets on Kubernetes, this error is not from authentication, the connection is not refused, there is any way to solve it?

Thanks for all you help!


EDIT: When I make a request to the server I get this error:

2017-06-02T02:09:51.417407254Z Request: GET /anime/1
2017-06-02T02:09:51.417410146Z ** (exit) an exception was raised:
2017-06-02T02:09:51.417413126Z ** (DBConnection.ConnectionError) connection not available because of disconnection
2017-06-02T02:09:51.417416219Z (db_connection) lib/db_connection.ex:926: DBConnection.checkout/2
2017-06-02T02:09:51.417421994Z (db_connection) lib/db_connection.ex:742:
2017-06-02T02:09:51.417425202Z (db_connection) lib/db_connection.ex:584: DBConnection.prepare_execute/4
2017-06-02T02:09:51.417429314Z (ecto) lib/ecto/adapters/postgres/connection.ex:80: Ecto.Adapters.Postgres.Connection.prepare_execute/5
2017-06-02T02:09:51.41743256Z (ecto) lib/ecto/adapters/sql.ex:243: Ecto.Adapters.SQL.sql_call/6
2017-06-02T02:09:51.417435814Z (ecto) lib/ecto/adapters/sql.ex:431: Ecto.Adapters.SQL.execute_and_cache/7
2017-06-02T02:09:51.417438957Z (ecto) lib/ecto/repo/queryable.ex:130: Ecto.Repo.Queryable.execute/5
2017-06-02T02:09:51.417442142Z (ecto) lib/ecto/repo/queryable.ex:35: Ecto.Repo.Queryable.all/4

Hey, which hostname are you using to connect to it? The error you’ve received shows a timeout, so it’s unable to connect to the host. It’s probably unrelated to Ecto or Elixir.


It’s connecting to a Google SQL instance, it already has access by ip and api, I don’t know why it don’t connect…

I already have check the secrets on the app, and its correctly too, also as mention isnt a authentication error :C

Hey, it’s not an authentication error indeed. Are you sure your Google Cloud SQL instance is Postgres, not MySQL? Also, make sure you’re using the right port.

Yeah I’m sure its postgres look:

This is all log/sql and config

You might need to add the server’s IP to the allowed IPs like here.

Its already added :confused:

I’d the same kind of issues some time ago with a node project. It really started happening randomly and what I did to fix it was an upgrade on gce-proxy image I use to access Cloud SQL form 1.05 to 1.09.

I really don’t know why it started to happen, and don’t know also what is the root cause of the error. I’ll investigate it soon.

1 Like

Have you tried using an instance of Google Cloud SQL proxy inside your deployment? so you can connect locally and authenticating using a Service Account credentials.

I don’t know if you are running Kubernetes inside GKE, but that’s the recommended way in that case. More info can be found on:

Hope that helps!

1 Like

I am getting the same error in Openshift (Kubernetes too).

As configuration in Kubernetes is done during deployment, it should be working using this configuration (and having the proper environment variables):

config :reservation_book, ReservationBook.Repo,
  adapter: Ecto.Adapters.Postgres,
  username: {:system, "DB_USER"},
  password: {:system, "DB_PASSWORD"},
  database: {:system, "DB_NAME"},
  hostname: {:system, "DB_HOST"},
  pool_size: 15

However, Ecto doesn’t seem to recognize the format {:system, “VARIABLE”} for all values, and thus I am stuck.

Perhaps the Ecto.Repo.init/2-callback can help you?

A short example of how to use it, can be seen in this post: