Crazy amount of guests

Screenshot 2026-11-19 082013.webp

Just took a look at some of the stuff going on right now, that 260k is a block of cf.client.bot and the logs show FB and Tencent / Byteplus (Tiktok), are the problem children. On my managed challenge, China seems to be buying their way through, along with some othe actors.
 
Just took a look at some of the stuff going on right now, that 260k is a block of cf.client.bot and the logs show FB and Tencent / Byteplus (Tiktok), are the problem children. On my managed challenge, China seems to be buying their way through, along with some othe actors.
I have been blocking Tencent Cloud and Chinanet for years on the firewall. Zero reason to have any traffic from them. I wish I could block Saleforce too, they are terrible for spam but some legit companies use them.

A couple of years ago Saleforce Blacklisted our email server because I was reporting messages as spam. Had to contact the corporate lawyer at Salesforce to get unblocked and they claimed they thought I want to opt out of messages from their network. SMH. Instead of blocking the spammers, they blocked me.


Blocked CHINA Tencent Cloud 42.192.0.0/15, 1.12.0.0/14, 43.136.0.0/16, 111.230.0.0/15, 43.156.0.0/16, 146.56.192.0/18, 43.135.128.0/18, 119.45.0.0/16, 162.14.0.0/16, 129.28.0.0/16, 125.88.0.0/13, 43.138.0.0/15, 175.27.0.0/16, 43.134.64.0/18, 119.28.120.0/22, 119.28.96.0/20, 119.28.112.0/21, 129.226.0.0/16, 43.134.0.0/18, 150.109.0.0/19, 43.163.193.0/24, 43.163.208.0/20, 43.163.192.255, 43.163.96.0/22, 43.163.200.0/21, 43.163.224.0/19, 43.163.194.0/23, 43.153.192.0/18, 43.130.224.0/19, 43.132.96.0/19, 129.226.64.0/20, 43.166.224.0/19, 43.128.128.0/19, 43.160.0.0/12, 43.153.128.0/18, 43.156.35.204, 43.153.0.0/17, 43.155.128.0/17, 121.4.0.0/15, 129.226.160.0/19, 101.32.0.0/18, 101.32.64.0/20, 170.106.0.0/17, 170.106.128.0/19, 170.106.160.0/21, 43.134.224.0/19, 43.154.0.0/16, 43.155.0.0/17, 43.128.64.0/18, 43.133.160.0/19, 124.156.96.0/19, 124.156.128.0/18, 43.157.128.0/18, 139.155.0.0/16, 101.33.64.0/19, 101.33.96.0/22, 43.157.192.0/18, 150.109.224.0/19, 150.109.208.0/20, 43.163.0.0/17, 43.132.128.0/18, 129.226.128.0/20, 129.226.96.0/19, 119.28.0.0/15, 43.133.32.0/19, 124.156.192.0/20, 43.133.96.0/19, 43.134.192.0/19, 43.157.0.0/17, 150.109.144.0/20, 161.120.168.0/22, 43.133.224.0/19, 162.62.208.0/20

Blocked ChinaNet & China Unicom & China Mobile & Cloudie 125.64.0.0/13, 119.120.0.0/13, 112.224.0.0/11, 114.224.0.0/12, 219.136.0.0/15, 219.128.0.0/13, 113.240.0.0/13, 116.224.0.0/12, 183.192.0.0/10, 113.112.0.0/13, 118.112.0.0/13, 182.32.0.0/12, 43.137.0.0/16, 180.152.0.0/13, 14.112.0.0/12, 222.72.0.0/15, 222.64.0.0/13, 118.195.128.0/17, 43.136.0.0/16, 182.128.0.0/12, 111.176.0.0/13, 180.96.0.0/11, 222.92.0.0/14, 183.160.0.0/13, 14.16.0.0/12, 210.16.128.0/18, 49.64.0.0/11, 111.0.0.0/10, 116.16.0.0/12, 120.192.0.0/10, 39.128.0.0/10, 183.0.0.0/10, 223.64.0.0/11, 223.112.0.0/14, 223.116.0.0/15, 223.96.0.0/12, 179.106.168.0/21, 168.194.208.0/22, 132.255.184.0/22, 105.242.0.0/16, 45.167.36.0/22, 191.241.192.0/19, 179.63.236.0/22, 203.130.32.0/19, 14.144.0.0/12, 117.24.0.0/13, 125.40.0.0/13, 101.44.160.0/20, 39.64.0.0/11, 117.50.0.0/16, 43.154.0.0/16, 43.153.128.0/18, 106.224.0.0/12, 182.84.0.0/14, 182.96.55.36, 27.16.0.0/12, 103.115.56.0/22, 103.115.48.0/22, 150.109.96.0/19, 36.227.0.0/16, 211.157.192.0/20, 211.157.128.0/18
 
Yer... but blocking means you can't see what they're up to. I like to understand what is going on so I can find ways to mitigate it precisely. Yes, you can block China, you can block the ASN, but they have servers worldwide they're using to scrape everything they can get their hands on. I want to understand their techniques to counter them correctly, and that way it doesn't matter what server they use, or from where in the world.
 
Yer... but blocking means you can't see what they're up to. I like to understand what is going on so I can find ways to mitigate it precisely. Yes, you can block China, you can block the ASN, but they have servers worldwide they're using to scrape everything they can get their hands on. I want to understand their techniques to counter them correctly, and that way it doesn't matter what server they use, or from where in the world.
Wouldn't that be like leaving your house unlocked so you can observe what the thieves like to steal once they are inside your home? :)
You can do both, as you said, they will use other methods anyway, it seems like the scale of the probes and scraping are lower via other servers they use.

I also use https://www.ipdeny.com to create firewall rules to block entire countries.

Gemini has been very helpful in creating abort rewrite rules to stop probes and scraping that are observed in the log files and http error logs.
 
I have been blocking Tencent Cloud and Chinanet for years on the firewall. Zero reason to have any traffic from them. I wish I could block Saleforce too, they are terrible for spam but some legit companies use them.

A couple of years ago Saleforce Blacklisted our email server because I was reporting messages as spam. Had to contact the corporate lawyer at Salesforce to get unblocked and they claimed they thought I want to opt out of messages from their network. SMH. Instead of blocking the spammers, they blocked me.


Blocked CHINA Tencent Cloud 42.192.0.0/15, 1.12.0.0/14, 43.136.0.0/16, 111.230.0.0/15, 43.156.0.0/16, 146.56.192.0/18, 43.135.128.0/18, 119.45.0.0/16, 162.14.0.0/16, 129.28.0.0/16, 125.88.0.0/13, 43.138.0.0/15, 175.27.0.0/16, 43.134.64.0/18, 119.28.120.0/22, 119.28.96.0/20, 119.28.112.0/21, 129.226.0.0/16, 43.134.0.0/18, 150.109.0.0/19, 43.163.193.0/24, 43.163.208.0/20, 43.163.192.255, 43.163.96.0/22, 43.163.200.0/21, 43.163.224.0/19, 43.163.194.0/23, 43.153.192.0/18, 43.130.224.0/19, 43.132.96.0/19, 129.226.64.0/20, 43.166.224.0/19, 43.128.128.0/19, 43.160.0.0/12, 43.153.128.0/18, 43.156.35.204, 43.153.0.0/17, 43.155.128.0/17, 121.4.0.0/15, 129.226.160.0/19, 101.32.0.0/18, 101.32.64.0/20, 170.106.0.0/17, 170.106.128.0/19, 170.106.160.0/21, 43.134.224.0/19, 43.154.0.0/16, 43.155.0.0/17, 43.128.64.0/18, 43.133.160.0/19, 124.156.96.0/19, 124.156.128.0/18, 43.157.128.0/18, 139.155.0.0/16, 101.33.64.0/19, 101.33.96.0/22, 43.157.192.0/18, 150.109.224.0/19, 150.109.208.0/20, 43.163.0.0/17, 43.132.128.0/18, 129.226.128.0/20, 129.226.96.0/19, 119.28.0.0/15, 43.133.32.0/19, 124.156.192.0/20, 43.133.96.0/19, 43.134.192.0/19, 43.157.0.0/17, 150.109.144.0/20, 161.120.168.0/22, 43.133.224.0/19, 162.62.208.0/20

Blocked ChinaNet & China Unicom & China Mobile & Cloudie 125.64.0.0/13, 119.120.0.0/13, 112.224.0.0/11, 114.224.0.0/12, 219.136.0.0/15, 219.128.0.0/13, 113.240.0.0/13, 116.224.0.0/12, 183.192.0.0/10, 113.112.0.0/13, 118.112.0.0/13, 182.32.0.0/12, 43.137.0.0/16, 180.152.0.0/13, 14.112.0.0/12, 222.72.0.0/15, 222.64.0.0/13, 118.195.128.0/17, 43.136.0.0/16, 182.128.0.0/12, 111.176.0.0/13, 180.96.0.0/11, 222.92.0.0/14, 183.160.0.0/13, 14.16.0.0/12, 210.16.128.0/18, 49.64.0.0/11, 111.0.0.0/10, 116.16.0.0/12, 120.192.0.0/10, 39.128.0.0/10, 183.0.0.0/10, 223.64.0.0/11, 223.112.0.0/14, 223.116.0.0/15, 223.96.0.0/12, 179.106.168.0/21, 168.194.208.0/22, 132.255.184.0/22, 105.242.0.0/16, 45.167.36.0/22, 191.241.192.0/19, 179.63.236.0/22, 203.130.32.0/19, 14.144.0.0/12, 117.24.0.0/13, 125.40.0.0/13, 101.44.160.0/20, 39.64.0.0/11, 117.50.0.0/16, 43.154.0.0/16, 43.153.128.0/18, 106.224.0.0/12, 182.84.0.0/14, 182.96.55.36, 27.16.0.0/12, 103.115.56.0/22, 103.115.48.0/22, 150.109.96.0/19, 36.227.0.0/16, 211.157.192.0/20, 211.157.128.0/18
I have China blocked as a country, plus Malaysia and Myanmar, and also ASN's for Tencent and Alibaba etc in other countries, - but why block all those IP's and not the ASN's? Or just block China as a country.

Singapore 139341 (Tencent), 136907, 141180, 151610 (Huawei) 134963 (Alibaba)
Thailand 137876 (Tencent again)

US 45102 (Alibaba cloud - global)
 
Wouldn't that be like leaving your house unlocked so you can observe what the thieves like to steal once they are inside your home?
Not really. One just wants to take what is already available online, they just don't want to respect whether they cripple your server or not. Anyone creating an account to steal hidden data, yep, that is the same, because now you've locked the door.

I don't agree with either, but I like to understand the problem to develop layers and shields. Just like home security, layers and shields to make it harder, so they hopefully move onto elsewhere that is easier.
 
Fortunately, our community’s audience is primarily focused in the US, Canada, and UK.
Depends. Given that the US are by far the biggest source of bots (as can be seen on CF radar) and loads of scrapers come from there, residential proxies as well as data-center based ones one might consider this rather unfortunate than fortunate. Definitively makes it more difficult to filter out the scrapers.
 
Can confirm - just yesterday I caught a few dozen bots trying to get into my private forum, hitting the same URL simultaneously through two different proxies in completely different parts of the world. Passed all Cloudflare filters for ASN/country, for dubious user-agent values and the captcha for not being authenticated; passed Bot Guard's filters as well.

At the speed this is relentlessly escalating, we might (hopefully) get some traction in two possible scenarios:
  • new RBLs gathering distributed intel over residential proxies and getting them blacklisted on the internet, maybe getting ISPs to react (not holding my breath on that one)
  • as this starts looking more and more like fully automated "hacking" by searching for credentials to abuse (for cryptomining or further botnet expansion) and exploiting vulnerabilities, some nations' security agencies might start to get involved, and these guys would not tread lightly.
Who had something on the note of "fighting for the survival of Internet" among their "where do you see yourself in 5 years" answers 4 years ago? :)
Another finding, and this is one should come at the surprise of no one: these bot runners are now using Google's Quad8 and Cloudflare's 1x4 public DNS. No longer using their ISP / datacenter provided DNS or locally hosted DNS. So DNS as a method of early-warning is effectively useless now unless you're those public DNS providers, being able to see the true query-source IP. :(

Back in the day, if an ISP's IP space ended up on Sorbs or Spamhaus, people/network admins were fairly quick to fix up the bad actors or habits. Now a days, getting massive uptake for a RBL or RBL-like list for abusive-proxies/residential-proxies will be laughed at by just about everyone. No one really seems to care about the residential proxy issue. Granted, they'll probably care about it a bit more when these Residential Proxies get used for DDOS and defrauding operations though!

We're likely at the point of the internet's wild west 'nature' meeting AI's 'aggressive mechanisms' with the automation combo. Automated probing has been a thing with AI (and prior to AI) for a decent while. While anecdotal, the rate of attacks that my servers have been taking in. What I am starting to see is something very new: AI-backed drones probing servers for 0-day attack vectors, if it returns data the probe is looking for, the AI-backed probe slams the server with additional attack vectors in seconds - no longer a delayed "we'll come back to this particular IP and machine later" mindset. I am also seeing blind-probe attacks hitting much faster and much more rapid than years past. I am talking about seeing upwards of 2,000 attacks and probing-type attacks per second from a single IP before reactive log monitoring gets around to scanning and blocking the traffic at the server level.

The only issue I noticed is that it would block Bunny CDN sometimes. That is why the script cross checks the Bunny List- https://api.bunny.net/system/edgeserverlist/plain

As for people using a VPN, VPN is kind of like the 1990s blocking your caller ID, nobody would answer. There is no reason for someone to be on our websites using a VPN, especially for e-commerce or forum registration. Nearly always associated with something nefarious.
Oh no, I'm not referring to external/WWW-based VPNs. I'm talking about routing via private IP space for internal communications, and in many instances via a VPN route. For example, I could be using 10.10.0.1/24 on top of the delegated IPv4 and IPv6 WWW/WAN addressing for all ingest servers, and then those same ingest servers having a route active that directs external client requests over that 10.10.0.1/24 network to an internal server group running on 10.20.0.1/24. Similar to how AWS does it with their Overlay Encapsulation and VPC with RFC1918 configurations. That level 1 list would more or less block those communications on a generic firewall configuration. :)

Maybe there is hope? According to this the AI bubble has burst in China.

Nah, no chance. Even if the AI bubble were to completely 'pop', the infrastructure for these botnets / scrapers using AI-enhancement or AI-backed will still exist, and these sort of issues will persist.
 
I used managed hosting myself a long time ago, and having root access was so freeing, although when I switched over it meant a few sleepless nights, worrying about getting hacked into. (A large forum I took over for was hacked at one point, but that was the early 2000s.)

I eventually got everything so buttoned up (along with killing off many sites that weren't being used) that I don't need to worry about it much. Like having two firewalls--the host has one, as do my servers, so if for some reason my rules ever fail I can fall back on theirs. At the moment I run Elasticsearch and MariaDB each on separate servers, and only access them via the local network at the host--the only IPs open to the public are 1) my own IP address and 2) the IP address of a server at a different host, so I always have a backdoor.
Now a days thankfully, with modernized web server services (sounds silly to write it like that), containerizing web services and locking down their cross-talking capabilities against the other containers/services is a good choice. Especially when deployments are managed via a version control system. Makes the "LOL WE HACKED U!!!!" skiddies quite upset when one can just revert and lock down services while a fix is deployed from upstream. Virtualization of servers has been nice as well... but it seems most folks want to go the route of slapping everything to provisionless-containers these days and not deal with virtualization 'stuff'.

Now, as with self-managed servers and having full root on the machine, be it VPS/virtualized or physical... once proper general firewalling is performed, hardening of services and on-server access limitations/ACLs, with some automation of log monitoring-->actions taken for bad actors, restless nights start to become a thing of the past...mostly. :D

The only thing that keeps me up at night is when a ZFS pool craps out due to a failing drive... or when another damned CVE 9+ exploit is posted. 2026 is leaning heavy on that CVE side. While I am glad rapid updates are being deployed by the maintainers of various projects, it is the sheer rate of these attack-vectors being discovered... that is mindboggling.
 
I had a small hit that got through yesterday - 165. This is because I had opened things up a bit for real users from one country so less get challenged. The 165 got absolutely nothing because there were only two pathways they could go to, to avoid the rules and neither of those give them anything. On the other hand if that 165 turned into 5,000 they would still go to the server, but I can't see why they would bother when they're not being able to scrape anything. They disappeared after a few minutes. I can always turn the challenge back on if needed - these aren't sophisticated ones.

I guess they are programmed to try one pathway after another until something gets through - mainly these were hitting member profiles where they get blocked at server level. The sophisticated ones are all still block/trapped.
 
Last edited:
but why block all those IP's and not the ASN's? Or just block China as a country.
Some of these strategies have been in place for years. Glad you brought this up, because I researched it and our datacenter rolled out new tools August 2025 that gives us more control over this! There is so much to keep up with I didn't even realize it was an option! Now blocking by Country/ASN.

It keeps protected traffic inside the datacenter network so attacks can be inspected, filtered, and stopped with lower latency and better operational visibility.
-Always-on filtering
-Custom rule control
-Built for low latency
 
For fellow server administrators here, has anyone contemplated if the following would be an efficient use of time (read: probably educationally good, but full waste of time):

With proxies, especially residential proxies, there is inherent RTT/latency increments due to the 1>2>3>4 style (botrunner source -> proxy-to -> server -> proxy-from and back to bot runner) of relay-like routing due to the proxy. A normal client, without a proxy/vpn is like a 1 -> 2 style (client -> server). Which got me thinking on this...

Nginx (and probably Apache2 - not used it in years!) has built in support for the $tcpinfo_ metrics ($tcpinfo_rtt, $tcpinfo_rttvar, $tcpinfo_snd_cwnd, $tcpinfo_rcv_space). By definition of what a proxy is and does, this would designate that there is a physical, and thus a measurable latency gap between the local TCP handshake and the tunneled TLS/HTTP payload.

At the TCP level, the handshake (SYN -> ACK) will terminate directly at the home router/device. Unless I am mistaking this, $tcpinfo_rtt is fast and local, to which I'd estimate anywhere between 15ms to 45ms.

Next, after the TCP Handshake level, we move into the TLS aspect, starting at the ClientHello and HTTP request into the web server. This traffic must travel back through the proxy provider's gateway (relay? I dunno - semantics at this point) to the attacker's machine and back. Which in theory, would potentially add anywhere from 80 to 500ms+ of "unexplained" but additional delay on a low-RTT connection.

Think this would be something viable to explore? This almost sounds too good to be true, but a proxy is indeed a proxy, and a proxy adds latency.
 
Coming soon in XenForo 2.3.14 are some improvements I think will be welcome. More info and discussion here:

 
I am very much looking forward to the above!

On another note, I've gone ahead and just blocked all of Chrome/154 and Chrome/150 altogether, with a message letting real users know how to contact us if they were caught in the ban. Been a few hours, no messages yet. But our Google Analytics are a lot cleaner so far - no more direct traffic being 3x the amount of search.
 
Coming soon in XenForo 2.3.14 are some improvements I think will be welcome. More info and discussion here:

Excellent - thank you :-)
 
I am very much looking forward to the above!

On another note, I've gone ahead and just blocked all of Chrome/154 and Chrome/150 altogether, with a message letting real users know how to contact us if they were caught in the ban. Been a few hours, no messages yet. But our Google Analytics are a lot cleaner so far - no more direct traffic being 3x the amount of search.
One option, if you're using Cloudflare, would be to set a custom rule to challenge all computers/desktops. Rather than block the browsers, as they keep switching browsers when blocked. The old Chrome ones can't solve Cloudflare challenges IME. Real guests would then just get a Cloudflare challenge instead of a block and members could be whitelisted so they get no challenge.

But depends if your site has more computer users than mobile users. Mine is mainly mobile users, so it's not that many real guests to challenge. I also whitelisted certain countries from it.
 
Last edited:
One option, if you're using Cloudflare, would be to set a custom rule to challenge all computers/desktops. Rather than block the browsers, as they keep switching browsers when blocked. The old Chrome ones can't solve Cloudflare challenges IME. Real guests would then just get a Cloudflare challenge instead of a block and members could be whitelisted so they get no challenge.

But depends if your site has more computer users than mobile users. Mine is mainly mobile users, so it's not that many real guests to challenge. I also whitelisted certain countries from it.
We have about a 50/50 mix of desktop/mobile and around 1m pageviews/month, so it'd be a substantial amount of people to challenge and would probably affect us negatively I think :(
 
Back
Top Bottom