Automating GIT workflow processes / Linux

Deployment process and architecture

For the development team needs it’s should to create custom Git repos architecture under Linux (CentOS 7) which should cover and automate the whole internal workflow and no manual tasks should be performed to keep the process up and maintenance the repos. The main target is every developer to have all program code to run the application by himself on his own working station but to have everything encrypted instead of the parts of the code /components/ they are working over. Git submodules doesn’t make any sense.

Repository architecture:

Git Main Repo Represents the main repo where only super-users have access to. The application source code in the main repository is complete and unencrypted. There are limited users who have access to this repository, called super-users and no one developers have

any access to.

Write and read to the main repos is available only for the super-users and the software

for

Dev’s 1,2,3…N repositories

Represents separate repositories for every developer of the team where different

parts(components) of the codes are accessible for the certain developer and the rest of

the code is initially encrypted so the developer has access only to his own components

and no any access to unencrypted components which he is not allowed. In the same

time all application code should be there and

Configuration

Store all information about main repos, dev’s repos, user permissions to different

components of the code in each repo

o Main definition of all users and their roles /super-user or developer/

Definition of their local and remote access data

o Definition of all repositories – end-developer repos or main-repos and their

owners

Definition of restricted (encrypted) components (by path) for each main

repository and its branches

Definition of allowed access for each developer to every component of

the project

Process automation

1. Main to dev repository synchronization

1.1. The code from main repository should be synchronized to all Dev’s repositories initially

and on every commit to the all developers with access to this project

• All code where developer doesn’t have access will be encrypted

• All commits from other users for parts of the code where current developer

doesn’t have access should be not visible

• All other commits and their comments to the part of code / components where no

restriction is applied should be visible to the user which owes current repository

1.2. Manual synchronization option

In any case there should be option for manual synchronization available to super users

1.3. Data adding, cleanup and replacing

For security reasons there should be options for excluding, including or replacing files

or information within files while synchronization is done.

It could be license files, files with sensitive information or settings for compatibility

2. Developer to main repository synchronization

2.1. All push requests and comments to the developer mirrored repo should be replicated

to the main repo from the same user ID

• Result from that request applied on the main repository should be also replicated

back to the developer repo

• Conflict should be solved on the developers repository if any

• Manual option to catch the conflicts to the level of main repository

3. Timeline

All changes and repo updates should have no more than 1 minute latency between

mirroring to all available repositories in relationship

4. Code encryption

Encryption of the code should be 100% automated and no manual work should be

performed.

Since used language is PHP encryption system will be IONCUBE PHP Encoder 10 which

allows command-line commands and automation for deployment

5. Automated deployment to live environment

Linux Java Ubuntu Programação C Rede de Administração

ID do Projeto: #26786098

Sobre o projeto