Did recent fly.io changes break docker/laravel connections with postgresdb?

Questions / Help
1

I have 2 fly.io services, dockerfile/laravel API and postgresdb. Both were working fine until a few days ago, but now my frontend API cannot query the DB. With this error.

2022-07-20 app[42a198b4] sin [info] [20-Jul-2022] WARNING: [pool www] child 4382 said into stderr: "[2022-07-20] production.ERROR: SQLSTATE[08006] [7] could not translate host name "flyio-db.internal" to address: Name does not resolve (SQL: select * from "users" where "email" = example@example.com limit 1) {"exception":"[object] (Illuminate\\Database\\QueryException(code: 7): SQLSTATE[08006] [7] could not translate host name \"flyio-db.internal\" to address: Name does not resolve (SQL: select * from \"users\" where \"email\" = example@example.com limit 1) at /var/www/html/vendor/laravel/framework/src/Illuminate/Database/Connection.php:716)"

The app and the db haven’t been touched in months. The app connects to the db via a Database URL string, which, again, was set months ago and not touched.

Has anyone documented any similar errors?

2

From that error message it sounds like it is not specific to Laravel or Postgres, but due to the private .internal DNS record not resolving. And hence anything relying on that domain (in your case a database query) would fail.

I wonder if this issue has resurfaced? :thinking: It certainly sounds similar:

3

Hmmm, that sounds like an awfully annoying problem. Will try to restart the app and see if it fixes itself like that question you sent.

Too bad I can’t restart the app via the dashboard, that would be handy.

Edit
Restarting the Laravel API app didn’t work. Restarting the DB worked.

Why on earth did fly db randomly stop recognising the db internal name? Not fun to have a deployment service that might randomly break shit.