haproxy configuration for using with prerender.io

haproxy.cfg

# Change YOUR_TOKEN to your prerender token
# Change http://example.com (server_name) to your website url

frontend my-frontend
    mode http
    bind :80
    
    # prerender.io
    acl user-agent-bot hdr_sub(User-Agent) -i baiduspider twitterbot facebookexternalhit rogerbot linkedinbot embedly showyoubot outbrain pinterest slackbot vkShare W3C_Validator
    acl url-asset path_end js css xml less png jpg jpeg gif pdf doc txt ico rss zip mp3 rar exe wmv doc avi ppt mpg mpeg tif wav mov psd ai xls mp4 m4a swf dat dmg iso flv m4v torrent ttf woff
    acl url-escaped-fragment url_sub _escaped_fragment_

    use_backend prerender if user-agent-bot !url-asset
    use_backend prerender if url-escaped-fragment !url-asset

backend prerender
   mode http
   timeout server 20s
   server prerender service.prerender.io:443 check ssl verify none
   http-request set-header X-Prerender-Token YOUR_TOKEN
   reqrep ^([^\ ]*)\ /(.*)$ \1\ /http://example.com/\2

Yeah! thank so much! you save my time!

I'm planning on running prerender locally. Obviously, I'd change the server line to something like:

server prerender localhost:3000 check ssl verify none

But what is the last line supposed to be doing? I'm not sure what the "example.com" is supposed to be.

A few corrections. Each extension needs to have a leading dot, otherwise anything with those trailing characters but not as an extension won't be prerendered (such as pagecss).

acl url-asset path_end .js .css .xml .less .png .jpg .jpeg .gif .pdf .doc .txt .ico .rss .zip .mp3 .rar .exe .wmv .doc .avi .ppt .mpg .mpeg .tif .wav .mov .psd .ai .xls .mp4 .m4a .swf .dat .dmg .iso .flv .m4v .torrent .ttf .woff

Also instead of using reqrep, it's better to use http-request set-path. Allows variables to be used, etc.

backend prerender
  server pre1 ${PRERENDER_HOST} check
  acl with_ssl ssl_fc
  http-request set-path /https://%[hdr(host)]%[path] if with_ssl
  http-request set-path /http://%[hdr(host)]%[path] unless with_ssl

It may actually be better to whitelist extensions rather than have a huge blacklist. Since the only pages that need to be prerendered are html pages, I think it would be safe to assume only .html, .htm, and no extension at all. This should do the trick:

acl url-asset path_reg ^(.*\.(html?)$)?[^.]*$

use_backend prerender if user-agent-bot url-asset
use_backend prerender if url-escaped-fragment url-asset

Saved me lots of time! Thanks! 👍

Since Google now supports dynamic rendering, the googlebot agent should be added to the user-agent-bot acl

acl user-agent-bot hdr_sub(User-Agent) -i googlebot baiduspider twitterbot facebookexternalhit rogerbot linkedinbot embedly showyoubot outbrain pinterest slackbot vkShare W3C_Validator

For those looking at migrating over to HAProxy 2.1 and above, the configuration file below is what we're using in production. It contains a few crucial additions such as the resolvers dns section to keep HAProoxy happy. Cheers!

# Important: Prerender's IP address changes regularly. This ensures HAProxy always resolves the new IP.
resolvers dns
    parse-resolv-conf
    hold valid 10s

frontend wwoof
    mode http
    bind *:80

    # Detect bot crawlers looking for pre-rendered pages
    acl user-agent-bot hdr_sub(User-Agent) -i googlebot bingbot baiduspider twitterbot facebookexternalhit rogerbot linkedinbot embedly showyoubot outbrain pinterest slackbot whatsapp vkShare W3C_Validator
    acl url-asset path_end js css xml less png jpg jpeg gif pdf doc txt ico

    use_backend prerender if user-agent-bot !url-asset

backend prerender
    mode http
    server prerender service.prerender.io:443 resolvers dns check ssl verify none # Note the "resolvers dns" here
    http-request set-header X-Prerender-Token YOUR_TOKEN
    http-request set-uri /http://example.com%[path] # set-uri will remove the query string, use set-path if you want to keep it