Resolved nginx/apache proxy issue on all domains and webmail / ubuntu 22.04 on arm / external ip instead of localhost

[themaxx32000]

Server operating system version

Ubuntu 22.04

Plesk version and microupdate number

18.0.60

- Ubuntu 22.04 on ARM
- Fresh install with restored backup on a new machine (new IP)

- all Websites and their Webmail-Subdomains throw 502`s by nginx
- disabling nginx-proxying fixes the Websites but not webmail (as it is always using apache)
- proxy_error_log shows connect() failed (111: Connection refused) while connecting to upstream, client for each request
note: each request is sent from nginx to apache using the servers external ip, i.e. https://w.x.y.z:7081 and not https://127.0.0.1:7081
- using curl via ssh locally on

https://w.x.y.z:7081

throws a curl: (7) Failed to connect to W.X.Y.Z port 7081 after 0 ms: Connection refused
- checking apaches listening ports shows that it is listening only on 127.0.0.1:7081 and 127.0.0.1:7080
- modifying /etc/nginx/plesk.conf.d/webmails/sample.com_webmail.conf 's proxy_pass rule from the servers external ip to 127.0.0.1 also fixes webmail

What is wrong here?

 

[themaxx32000]

more structured as I cannot edit my first message anymore:

Info:

• Ubuntu 22.04 on ARM
• fresh install with restored backup on a new machine (new IP)

Facts:

• all Websites and their Webmail-Subdomains throw 502`s by nginx
• disabling nginx-proxying fixes the Websites but not webmail (as it is always using apache)
• proxy_error_log shows connect() failed (111: Connection refused) while connecting to upstream, client for each request
• note: each request is sent from nginx to apache using the servers external ip, i.e. https://w.x.y.z:7081 and not https://127.0.0.1:7081
• using curl via ssh locally on w.x.y.z:7081 throws a curl: (7) Failed to connect to W.X.Y.Z port 7081 after 0 ms: Connection refused (same for w.x.y.z:7080)
• using curl via ssh locally on 127.0.0.1:7081 and 127.0.0.1:7080 works fine
• checking apaches listening ports shows that it is listening only on 127.0.0.1:7081 and 127.0.0.1:7080
• modifying /etc/nginx/plesk.conf.d/webmails/sample.com_webmail.conf 's proxy_pass rule from the servers external ip to 127.0.0.1 also fixes webmail

Questions:

• why is the servers external ip even in all of apaches config-files when apache is listening to localhost only?
• why is nginx using the servers external in its proxy_pass rules for the same reason?
• what is wrong here?

 

[themaxx32000]

on another x86 Plesk Server running 18.0.59, apache is listening on 0.0.0.0:7080 / 0.0.0.0:7081 instead of localhost / 127.0.0.1 (which of course would also fix the issue, however the configuration on the new server is different and I need to know how to fix this).

 

[themaxx32000]

alright, one step further: https://support.plesk.com/hc/en-us/...che-Localhost-Mode-in-Plesk-18-0-56-PPP-43250

with Apache Localhost Mode introduced in Plesk 18.0.56 apache is set to listen to localhost only. thats fine and good but this must then also change all nginx-proxy_pass-configs to reflect this change..

new question: how can I force Plesk to reflect those settings and re-apply it? plesk repair all does not fix this.

 

[themaxx32000]

ok I fixed the issue but this needs to be implemeted in the restore routine:

disabling and re-enabling the Localhost Mode fixes the configs:

plesk bin apache --listen-on-localhost false && plesk bin apache --listen-on-localhost true

 

[Kaspar]

Thank you for sharing the solution. From which Plesk version was the restored backup?

 

[themaxx32000]

Can't tell for sure. The server, the backup was taken from, went offline yesterday night, so I guess it would be either the latest .60 or .59

 

[Nygård]

This was the solution to my issue regarding "proxy mode" activated, crashing all domains on my server after recovering plesk from backup.
With the solution above, I am now able to activate "Proxy mode" again.
So simple!

 

[Vertex292]

ok I fixed the issue but this needs to be implemeted in the restore routine:

disabling and re-enabling the Localhost Mode fixes the configs:

plesk bin apache --listen-on-localhost false && plesk bin apache --listen-on-localhost true

This was the solution for me too after restoring my plesk installation from a backup generated on 18.0.61 and recovered on 18.0.61. All of my domains were crashing with 502 (unless I disabled proxy-mode, which is not a fix, obviously). Thank you!

 

[propz]

ok I fixed the issue but this needs to be implemeted in the restore routine:

disabling and re-enabling the Localhost Mode fixes the configs:

plesk bin apache --listen-on-localhost false && plesk bin apache --listen-on-localhost true

This was the solution for me too.

Similar background story:
- Fresh Plesk installation in Debian 12 (from Ubuntu 24)
- Same Server + Same IP
- Newly added IPv6 to the server
- Successfully restored Plesk from Backup (18.0.68)

ALL domains were crashing with 502 nginx Connection refused error.

Many thanks for figuring it out!