You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I've gotten my app and my database talking to each other, but my SQL and App Engine logs are being absolutely flooded by connection errors anyway.
On the App Engine side, I'm seeing the usual /cloudsql/project:region:database/.s.PSQL.5432 not found error. But on the Cloud SQL side, I'm seeing FATAL: remaining connection slots are reserved for non-replication superuser connections
Again: I have pages that launch DB queries that I can load reliably without 500s, so they are actually talking to each other, but something else is clearly going on here.
My limited research has shown that a base PostgresQL instance can support up to 100 connections, and the default pool size I've seen for Phoenix in production is 20.
The text was updated successfully, but these errors were encountered:
Access the production database from App Engine
The App Engine runtime also runs a Cloud SQL Proxy for you, and makes your databases available via Unix sockets. In the App Engine environment, these are located in the directory /cloudsql at the root of the file system. So, to prepare your app for deployment into App Engine, edit your config/prod.secret.exs file again, and modify the socket_dir database setting to point to the correct location for App Engine...
Note that if you need to run another Ecto migration or open another psql session from your local workstation, you can temporarily revert socket_dir to /tmp/cloudsql so that Phoenix can talk to your local Cloud SQL Proxy. If you do, make sure you change it back to /cloudsql before you deploy to App Engine. Alternatively, if you have the ability to create the directory /cloudsql on your local workstation, you can configure Cloud SQL Proxy to open its sockets there instead, and avoid the need to revert socket_dir.
I've gotten my app and my database talking to each other, but my SQL and App Engine logs are being absolutely flooded by connection errors anyway.
On the App Engine side, I'm seeing the usual
/cloudsql/project:region:database/.s.PSQL.5432
not found error. But on the Cloud SQL side, I'm seeingFATAL: remaining connection slots are reserved for non-replication superuser connections
Again: I have pages that launch DB queries that I can load reliably without 500s, so they are actually talking to each other, but something else is clearly going on here.
My limited research has shown that a base PostgresQL instance can support up to 100 connections, and the default pool size I've seen for Phoenix in production is 20.
The text was updated successfully, but these errors were encountered: