Rik Heywood's questions

I have a Node.js application that uses sockets.io to allow multiple users to edit the same document at the same time (a little like Google Docs).

I want to scale this out horizontally, but I'm stuck trying to figure out how to ensure all the users editing a specific document end up connected to the same Node.js instance.

For example, assume there are 2 documents (doc-A and doc-B), and two servers (server-1 and server-2). Someone connects and starts editing doc-A on server-1. Another user connects, also wanting to edit doc-A. They must be sent to server-1 to ensure the document remains consistent. Finally, a third user connects wanting to edit doc-B. We could send them to either server as no one is editing that document at the moment.

I'm sure every collaborative editing experience must solve this problem, but I can't figure out how to do it. Can anyone help?

I've disabled directory listings like so...

Options -Indexes

When I try and access a directory like this:-

www.example.com/secret/

I get a 403 Forbidden response.

However, I want a 404 Not Found response, so hackers can't guess my directory structure so easily. How would I do that?

I'm seeing a lot of notified-SplashPage and verify-SplashPage entries showing up in our apache error log

[Wed Jan 20 15:00:22 2010] [error] [client xxxxx] File does not exist: /home/site/htdocs/notified-SplashPage, referer: http://www.google.co.uk/search?hl=en&source=hp&q=xxxxx
[Wed Jan 20 15:00:22 2010] [error] [client xxxxx] File does not exist: /home/site/htdocs/verify-SplashPage, referer: http://www.google.co.uk/search?hl=en&source=hp&q=xxxxxx

There are others that come in a similar pattern (notified-Compliance etc) and always seem to be connected with google.

Anyone know what google is trying to find out? If it isn't google, does anyone know what they are meant to do?

We have Apache set up to gzip compress html pages before they are sent to the client browser.

However, some of our pages are slowish to generate and it seems that Apache is holding on until it has the complete page, compressing it, then sending it to the browser.

There are big chunks of the page (the main important bits) that are actually generated and output fairly quickly.

Is it possible to configure Apache to start compressing and send data for the page as soon as the script starts outputting something? Is it is, can you offer any help is how to do this?

If not, can you suggest any other way to get gzip compression working for the server?

The scripts that generate the pages are written in PHP. We are using Apache 2.0 on Linux.

I am based in the UK and need a new web server for a small ecommerce site. It is not big enough to need separate database and web servers, so everything will sit happily on a single machine.

I have been getting a few quotes, but thought I would see what people think is reasonable for a low to medium spec dedicated server (Linux)?

If you were ordering a new server today, what features would you want from the hardware and the hosting company?