Nginx CloudFlare IPv6 instead of Google?

• CentOS Version: CentOS 7 64bit
• Centmin Mod Version Installed: 124.000 Stable
• Nginx Version Installed: 1.21.6
• PHP Version Installed: 8.0.24
• MariaDB MySQL Version Installed: 10.3.36
• When was last time updated Centmin Mod code base ? : 10/08/2022
• Persistent Config:
  
  Code (Text):

  SET_DEFAULT_MYSQLCHARSET='utf8mb4'
  LETSENCRYPT_DETECT='y'
  DUALCERTS='y'
  SELFSIGNEDSSL_ECDSA='y'
  NGINX_LIBBROTLI='y'
  NGXDYNAMIC_BROTLI='y'
  PHP_BROTLI='y'
  PHP_LZFOUR='y'
  PHP_LZF='y'
  PHP_ZSTD='y'
  DMOTD_PHPCHECK='y'
  MM_LICENSE_KEY='xxxxxxxxxxxxx'
  MM_CSF_SRC='y'
  PUREFTPD_DISABLED='y'
  

Is there an issue with nginx not passing through the correct IPv6 addresses that come through CloudFlare that anyone else has noticed? I noticed that the Google 'bots refer back to Google IP's for IPv4... but the one Google bot using IPv6 came back to a CloudFlare IP

---

I have the following enabled in my site config

Code:

include /usr/local/nginx/conf/cloudflare.conf;

and it consists of the latest IP ranges, the specific one related being

Code:

set_real_ip_from 2a06:98c0::/29;

 

It's the standard process of it being reported to XF... so from what I understand, nginx should be passing the actual origination IP through due to the use of the "set_real_ip" string. I know prior to enabling CloudFlare on the site, the Google IPv6 addresses were reported correctly. So the only "new" thing that has been done is the include in the nginx config file and CloudFlare enabled. I'm wondering if their is an nginx bug involving IPv6 and passing through the actual IPv6?

And every one of the IP checks returned to CloudFlare, so nginx is passing through the CloudFlare IPv6 address instead of the Google one.

Since engaging CloudFlare, I'm getting some weird stuff elsewhere also. I can't upload an add-on through the ACP any longer, as I get a peer reset error (502 gateway cloudflare screen) in site error log:

Code:

2022/10/08 16:23:19 [error] 720#720: *195 recv() failed (104: Connection reset by peer) while reading response header from upstream, client: 64.188.192.81, server: astrowhat.com, request: "POST /admin.php?add-ons/install-from-archive HTTP/2.0", upstream: "fastcgi://127.0.0.1:9000", host: "astrowhat.com", referrer: "https://astrowhat.com/admin.php?add-ons/"

 

I think this is going to be something that is related to the BOT function I have enabled on CloudFlare.

The other issue with the error 4 [SigKill] is going to be directly related to the DO VPS I have set up and the CentMin install. It didn't happen on the older VPS, which was actually a different level and cost. This is one of the newer ones so there is apparently some difference there.
I finally got CloudFlare totally disabled and the issue was still there, with nginx giving a 502 error as PHP-FPM apparently takes a dump all over itself after the add-on installer upload the add-on and before it can extract it and begin it's process.
It's one of the "Basic" setups, not the "Premium" Intel/AMD. It's a simple 2vCPU/2GB setup.