Windows Production Apache Cluster. Mongrel sleeps

When running production mode on Windows 2003 using Apache and a proxy balancer mod and mod_rewrite to redirect to the mongrel instances, I sometimes get a Proxy Error thrown from Apache.

Setup is done according to this guide:

4 Mongrel processes run as Windows Services on high ports (4000+)

VHost entry:

<Proxy balancer://ruby_cluster>   BalancerMember   BalancerMember   BalancerMember   BalancerMember </Proxy>

<VirtualHost *:80>   ServerName ************   DocumentRoot C:/websites/***********/

  <Directory C:/websites/**********/public/ >     Options Indexes FollowSymLinks MultiViews     AllowOverride All     Order allow,deny     allow from all   </Directory>

  ErrorLog C:/websites/**********/log/apache_error.log   LogLevel warn

  CustomLog C:/websites/**********/log/apache_access.log combined

  RewriteEngine On

  # Check for maintenance file and redirect all requests   RewriteCond %{DOCUMENT_ROOT}/system/maintenance.html -f   RewriteCond %{SCRIPT_FILENAME} !maintenance.html   RewriteRule ^.*$ /system/maintenance.html [L]

  # Rewrite index to check for static   RewriteRule ^/$ /index.html [QSA]

  # Rewrite to check for Rails cached page   RewriteRule ^([^.]+)$ $1.html [QSA]

  # Redirect all non-static requests to cluster   RewriteCond %{DOCUMENT_ROOT}/%{REQUEST_FILENAME} !-f   RewriteRule ^/(.*)$ balancer://ruby_cluster%{REQUEST_URI} [P,QSA,L] </VirtualHost>

Now this all works fine. Apache gets hit on port 80 like normal and redirects its request to one of the 4 mongrel services which will render the appropriate response.

However when the website hasn't seen traffic in a long while, lets say roughly 24 hours, the first hit always results in a 502 Proxy Error from apache:

Proxy Error

The proxy server received an invalid response from an upstream server. The proxy server could not handle the request GET /.

Reason: Error reading from remote server

Refreshing the page will display the ruby generated page correctly again.

It seems that Mongrel falls asleep or something like that and will react either too late or with an invalid response when queried by Apache.

Does anyone have any advice on how to counter this problem?