I have 5 web servers (Apache/mod_perl) behind Amazon EC2 Elastic Load Balancing, when I deploy codes to the web servers, I am doing this..

1. For each machine, shutdown the Apache
2. Update the code
3. Start over the server and proceed to the next server

I think when my server is shutdown, ELB will not distribute request to my server, but how about the request still serving?

I think a better approach is

1. Stop accepting new request from ELB
2. Sleep for sometimes, shutdown web server only if all requests are responded
3. Update the codes
4. Start the server again

But how to perform (1) and (2) from my local sever? Do I need to use AWS API? or other easy way to do it?

Thanks.