diff --git a/roles/pagure/templates/0_pagure.conf b/roles/pagure/templates/0_pagure.conf index ab47b7c4a3..8141188803 100644 --- a/roles/pagure/templates/0_pagure.conf +++ b/roles/pagure/templates/0_pagure.conf @@ -176,6 +176,21 @@ MaxConnectionsPerChild 1000 RewriteCond %{REQUEST_URI} ^/fork/mohanboddu/fedora-infra/ansible.*$ RewriteRule .* - [F] + RewriteCond %{REQUEST_URI} ^/fedora-docs/fedora-docs-web/blob.*$ + RewriteRule .* - [F] + + RewriteCond %{REQUEST_URI} ^fedora-qa/os-autoinst-distri-fedora/blob.*$ + RewriteRule .* - [F] + + RewriteCond %{REQUEST_URI} ^releng/fedora-scm-requests/blob.*$ + RewriteRule .* - [F] + + RewriteCond %{REQUEST_URI} ^pagure/usage/index.html.*$ + RewriteRule .* - [F] + + RewriteCond %{REQUEST_URI} ^/fork/.*/koji/.*$ + RewriteRule .* - [F] + # Spiders-gone-wild # These spiders may not follow robots.txt and will # hit admin sections which consume large amounts of CPU @@ -256,6 +271,13 @@ MaxConnectionsPerChild 1000 RedirectMatch ^/SSSD.sssd/index.html https://sssd.io/index.html RedirectMatch ^/SSSD.sssd/(.*) https://sssd.io/docs/$1 +# Spiders-gone-wild +# These spiders may not follow robots.txt and will +# hit admin sections which consume large amounts of CPU + RewriteEngine On + RewriteCond %{HTTP_USER_AGENT} "(Bytespider|ClaudeBot|Amazonbot|YandexBot|ChatGLM-Spider|GPTBot|Barkrowler|YisouSpider|MJ12bot)" [NC] + RewriteRule .* - [F,L] + WSGIProcessGroup paguredocs