An interesting finding about the blog and app* hosts

Hello everybody,

Yesterday night I discovered something interesting regarding our recurring database problem.

Although the fact that I was away from the computer last week. I was still making 'flush hosts' on both database servers. 

When I came back, I came to a realization. At every time a server had a problem, it was always app4 that was down. Another interesting fact is that we do not use appN to host the blog anymore, but the blog was still answering from the direct IP address. 

When we look at the apache log, we see a lot of repetitive requests, I think it would be a good experiment to do the following:
Default vhost (e.g. via the IP) to serve a static file
Each vhost only allows http communication to known hosts (i.e. fastly, aremysiteup, monitoring, etc)
Clean up the hosts with only the minimal web applications deployed with their vhosts
Use/configure fastly monitoring (I need to read about what we can get out of it)
Use fastly 50x error pages with static  
… among others...

I added them to the task list here [0] in the Refactor overview section [1]. 

As a result of my yesterday experiment, to remove wordpress code repository and apache vhost; I would have to 'flush host' twice a day still. But now, I still did not had to do any handling.

Regarding those tasks, I will look with Ryan and Doug for their priority level, analyze the time to solve them, and add them to the project management [2]

  [0]: http://docs.webplatform.org/wiki/WPD:Infrastructure/analysis/2013-Usage_and_future_state#3.2.2._Misc._tasks
  [1]: http://docs.webplatform.org/wiki/WPD:Infrastructure/analysis/2013-Usage_and_future_state#3.2._Refactor_overview
  [2]: http://project.webplatform.org/infrastructure/issues/open


Regards,

Renoir Boulanger  |  Developer operations engineer
W3C  |  Web Platform Project

http://w3.org/people/#renoirb  ✪  https://renoirboulanger.com/  ✪  @renoirb
~