When I thought about code/script organization it was important to make provision solution which could afford:

Use the same provision scripts for testing/development/production.
Make stand provisioning on-fly without additional artifact preparations
Use versioning (use specific versions on production stand, and development (sometimes SNAPSHOT) versions on development/stand environments)

Main tools

gradle - main build tool
docker-compose - for docker group deployment automation
docker - lightweight container for linux
puppet - provisioning tool

About deployment submodule structure

“docker_template” - Docker templates
“puppet_config” - puppet’s config files to change in container and docker images
“modules” - custom puppet modules for provisioning (infrastructure component, os-specific settings and so on)
“production” - production-specific script, Puppet’s Hiera databse and puppet’s sites.pp
“test” - test-specific script, Puppet’s Hiera databse and puppet’s sites.pp
“development” - development-specific script, Puppet’s Hiera databse and puppet’s sites.pp
“teststand” - test-specific script, Puppet’s Hiera databse and puppet’s sites.pp
provision_development.sh - provision-batch file
provision_production.sh - provision-batch file
provision_test.sh - provision_production.sh
provision_teststand.sh - provision_production.sh

Versioning

All versions should be specified in puppet’s config (on stand or production environment) in yaml %{environment}/hieradata/version.yaml, so you should edit necessary configs

About stands

General note

there are several kinds of stands:
- development - stand for development where all componenets should be run manually
- test - stand for testing purposes
- production - main production stand
stand initialization is done with help of gradle tasks (from store_line2_build repo “build_scripts/build-tasks.gradle”)
!!!names of containers should be the same as names of projects/submodules
all services are running under control and monitoring monit - it gives nice UI to see service status (see at Monitoring Links)

About “test”/”development”

stand preparation is done with help of gradle prepareStand (creates necessary files but doesn’t run anything)
real run is done with help of docker/docker-compose, so it must be done under root account

Sreparation steps:

Stand type detected (development by default) but could be changed with help of additional parameter, gradle -Pstand_type=test)
Configuration file story_line2_build/docker-%{environment}.properties is read
Directory for data, logs and stand dir is created
Docker templates from “docker_templates” are copied with substitution from story_line2_build/docker-%{environment}.properties
if version of component used “presented” completed artifact is copied to dir {docker_cont}/artifacts

Running steps (under root)

in stand dir docker-compose (docker-compose.yml), is ran (docker ‘ENTRYPOINT’) - a lot of different bash scripts could be used for it

Table of content: