> I tried to deploy new Rails app code to a production server the other
> day and took down the live site in production. I'd like to avoid this
> in the future. Something happened in the process that affected the
> live site. The error log showed that fastthread was not found. I had
> to go into the fastthread gem and run setup.rb. Then it told me that
> Rails was not installed so I had to do a gem install Rails. Then
> there were other gems that were needed by the application and not
> found so I had to reinstall those too.
> I have my own staging server that this all worked on, but the staging
> server is able to serve multiple sites so it is not an exact replicate
> of the client's production box. The Apache setup is such that only
> one site can be served from this machine - the document root in Apache
> is /project/public.
> I've got a couple of questions. 1- Any idea why gems would need to be
> reinstalled? 2- Can I deploy Capistrano without restarting Passenger
> to put the code onto the production server and then once it's all
> there, change the document root in Apache then do a deploy with the
> Passenger restart?
> Thanks,
> Cindy
Hi Cindy, did the client mix and match operating system-native packages and rubygems? That can cause problems, and it wouldn't surprise me if it caused this one.
I'm still kind of a noob with cap, so I'm not sure about question 2...
Have you thought about using virtualization to create real staging environments for your clients? It's a pretty fundamental flaw if your staging environment is not an exact replica of the production environment (save the staged code).
|
