On Thu, Dec 12, 2013 at 4:24 PM, Charles W Buege <cbu...@moreycorp.com>wrote:
> Periodically, we need to have our web sites down for maintenane - updates, > backups, etc. What I've been told to do is find a way to have a 'front > end' to our web sites so that, when they are down for maintenance, that > then end-users will receive a message 'This site is down for maintenance > till XX:XX AM/PM' or such. > How many sites (servers, vhosts)? How often is the downtime and is it predictable? > - What would you call what I am trying to setup? The reading I've done > implies that what I really want to setup is a reverse proxy server. Is > that what I want to do? > If you don't need a reverse proxy for any other reason, then it seems to me that it would be overkill to set one up just for this. > - If it isn't a reverse proxy I want to set up, what do I want to setup? > You could add a line to the config of each vhost that redirects all requests to a temporary error page. I have done this with a PHP script that returns a nice page with a 503 status code. RewriteEngine on RewriteRule !maintinence.php$ /maintinence.php [P] (Note: I am recreating this from memory because I can only log into the server from my office, so YMMV.) I put this commented out in a file that is included in every vhost, so I change it once and run httpd-k graceful > - Is this a type of 'clustering/high availabity' project I'm really > looking at here? I've been hearing those terms thrown about as well. > High-availability would be that you want to have more than one server so that if one stops working, the websites will still be available. Sometimes this can include taking one server offline to update it and then switching all traffic to that one while you update the second one. (Server combinations like nginx-unicorn for Ruby/Rails have that built in.) - Y
