[Spce-user] [EXTERNAL] Re: nginx and prometheus-nginx-exporter fail to load after reboot

[Cesar Mora]

I see. So this install is busted then… Having no web UI on restart and manually starting nginx every reboot is not reliable for a production server. I just wonder what causes this on a physical server only, but not on a VM server. Anyways, I’ll see what I can do with a script or something, or migrate the server to a VM…

Thanks!

> On Jan 27, 2025, at 7:32 AM, Michael Prokop via Spce-user <spce-user at lists.sipwise.com> wrote:
> 
> * Cesar Mora [Mon Jan 27, 2025 at 07:18:38AM -0400]:
> 
>> I’m so sorry! My email client keeps messing up the TO address.
>> 
>> I found a conflict of port 8081 with nginx and tproxy, not sure if this is a part of nginx.
>> 
>> root at spce:/# grep -Rn "tproxy" /etc/ngcp-config/
>> /etc/ngcp-config/templates/etc/services.tt2:339:tproxy          8081/tcp                        # Transparent Proxy
>> /etc/ngcp-config/templates/etc/services.tt2:383:#nginx-status                   [% nginx.status_port %]/tcp # Conflicts with tproxy
>> /etc/ngcp-config/templates/etc/kamailio/proxy/proxy.cfg.tt2:921:        # for BYE, mark whether or not to stop rtproxy
>> root at spce:/#
>> 
>> Further digging, I only found 2 references to port 8081 on /etc/nginx/
>> 
>> root at spce:~# grep -Rn "8081" /etc/nginx/
>> /etc/nginx/sites-enabled/nginx-status:2:        listen localhost:8081;
>> /etc/nginx/sites-available/nginx-status:2:      listen localhost:8081;
> 
> Ok, good and as expected, then.
> 
>> But I’m guessing the sites-enabled entry is a symlink
> 
> Yes, and only the files in /etc/nginx/sites-enabled/ matter for
> nginx in the end.
> 
>> Other than this, only tproxy refers to 8081 under the config.
> 
> The tproxy definition in /etc/services (and its template file
> /etc/ngcp-config/templates/etc/services.tt2) isn't relevant for
> this.
> 
> So this applies here:
> 
>>> But if there really is no other file referring to port 8081, this is
>>> about the issue that Guillem already pointed you to. quoting from
>>> his mail:
>>> 
>>> | This is a known limitation from this prometheus exporter, which fails
>>> | to start if nginx is not running. This has been fixed upstream with
>>> | https://github.com/nginx/nginx-prometheus-exporter/pull/504, so on
>>> | the future NGCP release where we rebase on top of Debian trixie,
>>> | we'll get that new fixed behavior.
> 
> regards
> -mika-
> 
> -- 
> Spce-user mailing list
> Spce-user at lists.sipwise.com <mailto:Spce-user at lists.sipwise.com>
> http://lists.sipwise.com/mailman/listinfo/spce-user_lists.sipwise.com

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.sipwise.com/pipermail/spce-user_lists.sipwise.com/attachments/20250127/4d7b5db5/attachment.htm>