Freshly provisioned and deployed environment not accepting connections

trellis
,
1

I’ve recently provisioned a new server and deployed essentially a boilerplate bedrock / sage10 setup to test things. At first the server responded with a 500 but I had yet to deploy. After the deploy it says it is refusing connections. I can ssh fine into the machine and see that the deploy was successful as the code is located in /srv/www/project.com/current/. I’m not sure what the issue is.

My only idea is that it is an issue with the wordpress_sites → local_path value from the environments group_vars wordpress_sites.yml.

The docs WordPress Sites | Trellis Docs | Roots say

  • local_path - path targeting Bedrock-based site directory (required)

This is set by default to ../site but I don’t see what that path could be referencing. I tried changing it to /srv/www/project.com/current/ and reprovisioning but this does not change anything.

I also checked the server firewall and connections are being allowed on both http/80 and https/443. I also rebooted the machine, but no change.

One note is that this is a custom environment. Meaning it is not staging.project.com but sage-10.project.com, but I don’t see that making a difference.

Thank you in advance for any tips.

2

How did you provision the server and how did you deploy to it? Was it the standard Trellis commands or did you use another method?

3

:thinking: It could also be caused by fail2ban (had a similar issue, with SSH connections). Have you attempted connections with a different IP address?

4

Standard Trellis commands. Both succeeded according to the logs.

Provisioning and deploying to an AWS Lightsail instance.

5

How do you mean? I’m trying to connect over http to the domain name. This is also how I’m connecting via SSH which works.

6

Often there is a cloud/hardware firewall in front of the virtual machine which also has to be configured to allow HTTP/HTTPS. SSH is often open by default, so it would work without explicit firewall configuration, but HTTP/HTTPS does not.

2 Likes
7

The firewall setting are the same as another Lightsail instance that responds without issue. Also the staging site.

I will destroy the Lightsail instance and recreate it tomorrow to see if perhaps I missed something.

1 Like
8

Some obvious tests:

  • netstat -taupen | grep nginx Is nginx actually listening on port 80 and/or 443?
  • service nginx status
  • wget http://thedomain.tld on the web server itself, does that work? Note that just using localhost normally does not work by nginx Trellis default configuration.
1 Like
9

Honestly you might just try destroying and reprovisioning the server.

10

For what it’s worth (nothing) I had the exact same issue the first time I tried deploying to an ARM AWS server for the first time. Second time it worked flawlessly though.

My best guess is that this is a super rare bug and that like me, you happened to stumble onto it.