Hi all,

A few hours ago, git.parabola.nu started getting slammed by requests,
and maxing out the CPU.  The requests were coming from changing IPs
(one IP might be used for 30ish requests) and randomized User-Agent
strings taken from actual browsers (though often quote old browser
versions).  Of the ~40,000 IPs that hit git.parabola.nu in a 3-hour
block, about a quarter of them are from AS136907 (Huawei Cloud).

I've gone ahead and set up iptables rules to block all of AS136907.
LMK if this causes any issues for legitimate users.

I'll watch `sudo iptables --verbose --numeric --list INPUT` over the
next few days and might relax it to just a few subnets.

For comparison with what we see in a few days, here's a `sort -n` of
the subnets that have tripped it so far:

     pkts bytes target     prot opt in     out     source               
destination         
        6   360 DROP       all  --  *      *       101.44.64.0/19       
0.0.0.0/0           
        6   360 DROP       all  --  *      *       119.8.24.0/21        
0.0.0.0/0           
        9   629 DROP       all  --  *      *       119.8.232.0/21       
0.0.0.0/0           
       12   720 DROP       all  --  *      *       119.13.64.0/18       
0.0.0.0/0           
       12   720 DROP       all  --  *      *       150.40.128.0/17      
0.0.0.0/0           
       12   720 DROP       all  --  *      *       182.160.0.0/19       
0.0.0.0/0           
       16   960 DROP       all  --  *      *       119.12.160.0/20      
0.0.0.0/0           
       18  1080 DROP       all  --  *      *       119.8.32.0/19        
0.0.0.0/0           
       18  1080 DROP       all  --  *      *       159.138.128.0/19     
0.0.0.0/0           
       18  1080 DROP       all  --  *      *       189.1.192.0/18       
0.0.0.0/0           
       24  1440 DROP       all  --  *      *       101.46.0.0/20        
0.0.0.0/0           
       24  1440 DROP       all  --  *      *       27.106.0.0/17        
0.0.0.0/0           
       30  1800 DROP       all  --  *      *       121.91.168.0/21      
0.0.0.0/0           
       30  1800 DROP       all  --  *      *       159.138.0.0/18       
0.0.0.0/0           
       48  2880 DROP       all  --  *      *       190.92.192.0/18      
0.0.0.0/0           
       52  3224 DROP       all  --  *      *       94.74.64.0/18        
0.0.0.0/0           
       54  3240 DROP       all  --  *      *       166.108.192.0/18     
0.0.0.0/0           
       60  3600 DROP       all  --  *      *       101.44.160.0/19      
0.0.0.0/0           
       61  3660 DROP       all  --  *      *       110.238.104.0/21     
0.0.0.0/0           
       66  3960 DROP       all  --  *      *       159.138.96.0/19      
0.0.0.0/0           
       82  4920 DROP       all  --  *      *       101.44.0.0/18        
0.0.0.0/0           
       84  5040 DROP       all  --  *      *       188.239.0.0/18       
0.0.0.0/0           
       91  5364 DROP       all  --  *      *       122.8.128.0/18       
0.0.0.0/0           
      107  6420 DROP       all  --  *      *       111.119.192.0/18     
0.0.0.0/0           
      150  9000 DROP       all  --  *      *       46.250.160.0/19      
0.0.0.0/0           
      164  9944 DROP       all  --  *      *       124.243.128.0/18     
0.0.0.0/0           
      533 27608 DROP       all  --  *      *       114.119.128.0/18     
0.0.0.0/0           
      936 56160 DROP       all  --  *      *       202.76.160.0/19      
0.0.0.0/0           
     3094  186K DROP       all  --  *      *       146.174.128.0/18     
0.0.0.0/0  

I'm not particularly keen on playing IP-wack-a-mole with
poorly-behaved AI scrapers.  How do you all feel about deploying
Anubis (https://anubis.techaro.lol/) in front of git.parabola.nu and
maybe other subdomains?  It's already used on gcc.gnu.org and a few
archlinux.org domains.

-- 
Happy hacking,
~ Luke T. Shumaker
_______________________________________________
Dev mailing list
[email protected]
https://lists.parabola.nu/mailman/listinfo/dev

Reply via email to