Some propositions to enhance the appliance deployment scripts for a better reusability

[syphax-bouazzouni]

Hi,

I helped lately the enit to deploy for them an appliance for the industry ontologies (https://www.enit.fr/en/index.html), and your deployment script worked very well when we use your repos. But when it comes to using our own repos/codes and config files it is more complicated.

So my feature request/proposal is to: Make the scripts more customizable and extract all the variables (github sources,, tags, the path to config files) in a single extracted config file (written in JSON, YAML or whatever)

Just for information also, at http://agroportal.lirmm.fr/ and http://bioportal.lirmm.fr/ we have our own deployment scripts, written completely using Capistrano/Ruby (no bash scripts) to :

• Have multiple environments ( agroportal, bioportal, stageportal and testportal)
• Update the code in servers (git pull)
• Bundle the code if ruby or maven build if java
• Update config files from an external and private repo containing all our config files

[syphax-bouazzouni]

And also others possible improvements, would be to add :

• a script to set/update the appliance ruby versions (using rbenv)
• a script to set/update the appliance java versions (using the centos alternatives command)
• scripts to set/update the httpd and Nginx config files
• scripts to install/update the dependencies (Redis , Memcached, mgrep, ...)
• have a script that gives us the current versions of all our components (code + dependencies) and tell if we are up to date with the components with the last version of the appliance.

The purpose of all this is to:

1. Make it easier for the forked version of your appliance to stay updated
2. Build an appliance from zero (from centos) without using docker
3. Make it more configurable, so that we can (the forked version) reuse your scripts.

Thanks,