Restarting Phoenix app gives me lots of Postgrex and DBConnection errors

Phoenix Forum Questions / Help
, ,
1

On some of my app deploys, I have pages of Postgrex and DBConnection errors.

Feb 28 09:06:23 postgrex Postgrex.Protocol 14:06:23.313 mfa=Postgrex.Protocol.handshake_shutdown/3 [error] Postgrex.Protocol (#PID<0.3358.0>) timed out because it was handshaking for longer than 15000ms 
Feb 28 09:06:23 db_connection DBConnection.Connection 14:06:23.314 mfa=DBConnection.Connection.handle_event/4 [error] Postgrex.Protocol (#PID<0.3358.0>) failed to connect: ** (DBConnection.ConnectionError) tcp recv (idle): closed 
Feb 28 09:06:23 postgrex Postgrex.Protocol 14:06:23.688 mfa=Postgrex.Protocol.handshake_shutdown/3 [error] Postgrex.Protocol (#PID<0.3360.0>) timed out because it was handshaking for longer than 15000ms 
Feb 28 09:06:23 db_connection DBConnection.Connection 14:06:23.690 mfa=DBConnection.Connection.handle_event/4 [error] Postgrex.Protocol (#PID<0.3360.0>) failed to connect: ** (DBConnection.ConnectionError) tcp recv (idle): closed 
Feb 28 09:06:23 postgrex Postgrex.Protocol 14:06:23.927 mfa=Postgrex.Protocol.handshake_shutdown/3 [error] Postgrex.Protocol (#PID<0.3357.0>) timed out because it was handshaking for longer than 15000ms 
Feb 28 09:06:23 db_connection DBConnection.Connection 14:06:23.929 mfa=DBConnection.Connection.handle_event/4 [error] Postgrex.Protocol (#PID<0.3357.0>) failed to connect: ** (DBConnection.ConnectionError) tcp recv (idle): closed 
Feb 28 09:06:24 postgrex Postgrex.Protocol 14:06:24.332 mfa=Postgrex.Protocol.handshake_shutdown/3 [error] Postgrex.Protocol (#PID<0.3363.0>) timed out because it was handshaking for longer than 15000ms 
Feb 28 09:06:24 db_connection DBConnection.Connection 14:06:24.334 mfa=DBConnection.Connection.handle_event/4 [error] Postgrex.Protocol (#PID<0.3363.0>) failed to connect: ** (DBConnection.ConnectionError) tcp recv (idle): closed

My database has plenty of connnections available, no errors, and when I try to connect to it manually it responds quickly.

How can I determine if this issue is with the startup of my new app, or teardown of my previous app? I’m deploying on Fly.io, targeting a single machine.

Thanks!

2

Well, it’s unlikely anything to do with your old application.

You can try something like fly m clone to setup another machine and see if it also has postgres connection errors. I’m thinking this might be something dns related, but there’s just not enough information here.

3

Yeah, that’s a good bet. I started getting a lot more of these after my DB moved from IPv4 to IPv6 and I changed the address accordingly. I found a single post about slow IPV6 connections on Fly.

I’m happy to add more info- I’m just not even sure where to start. This happens sporadically for my deploys, so even spinning up a new machine isn’t guaranteed to reproduce the issue.

4

I need a few more deploys to confirm, but it looks like this was an issue with DB pooling. I suspect on deploys I didn’t have enough spots for the shutting down and starting apps at the same time.

1 Like