[error] Postgrex.Protocol (#PID<...>) disconnected: ** (DBConnection.ConnectionError) tcp recv (idle): closed
I’m seeing intermittent ConnectionErrors on my unmanaged Postgres cluster. There are rather long quiet periods with no errors. But roughly a day or so apart (but sometimes as close as an hour apart), we’ll see 2 minutes of this error message on our Phoenix app servers.
Meanwhile, I don’t see any memory or CPU usage spikes on the DB machines at all. We’re way under our memory limit, and our CPU is under 3%. Not sure what else would be causing this because I’m not seeing any other errors in the logs. Perhaps it’s a networking issue? Any help on where to look would be appreciated.