Ci with jenkins docker and mssql belgium

CI Build Pipelines With Jenkins,
Docker and SQL Server
Chris Adkin

Who Do We Have Here Today ?
Some show of hands type questions

What Is Docker ? What Is Jenkins ?
 A continuous integration build
engine
 Written in Java
 Open Source
 Free !!! 
 Has its own conference
=> Jenkins World
 Virtual machines Vs Containers
 Containers and images
 Registries
 Pets and cattle
 The rise and rise of Docker . . .
 Orchestration

Why Use Jenkins ?
1. Plugin
support
Why Jenkins ?

Why Use Jenkins ?
2. Pipeline as code in two ‘Flavours’
Why Jenkins ?
Declarative pipeline syntax:
 Opinionated
 Highly readable
 Easy to learn
Scripted pipeline syntax:
 For ‘Power’ users
who want the
greatest degree of
flexibility
 Powerful, but harder
to learn than the
declarative pipeline
syntax

Why Use Jenkins ?
3. Docker pipeline
Why Jenkins ?

Why Use Jenkins ?
Why Jenkins ?
4. Jenkins shared libraries
=> enables code reuse across pipelines
5. Java based
=> runs in the same way wherever you can run Java
6. Light weight
=> the whole CI environment can be containerised
7. Community driven
=> evolves at pace
The
Best
Of the
Rest

Jenkins and The Public Cloud
 CloudBees => Jenkins-as-a-service
 Azure market place Jenkins vm image
 VSTS integration for Jenkins

What CI Value Do Containers Add ?
 Why use a full blown virtual or physical machine
for CI that is performed a few times a day ?
 Spin containers up on a free Linux distribution
 Great for performing pipeline steps in parallel
 Containers as build agents
 Entire build environment can be containerised

Demonstrations
May the demo gods
smile upon me this day !!!May
Demonstrations
. . . and finish here
A multi-branch build
pipeline using containers,
tSQLt, parallelism and
Docker volumes
We will start off simple here
A simple webhook triggered
build pipeline using a
non-containerised
SQL Server instance

1. Make a change to a SQL
Server Data Tools Project
2. Commit and push the
changes to a local GIT repo
3. A webhook
triggers the Jenkins
build
#1 A Simple Webhook Triggered Pipeline
GitHub repository

Points Of Interest In Demo #1
 A Git webhook
 Our first look at Jenkins-pipeline-as-code
 Configuring the msbuild plugin
 Use of stash to store the dacpac for later use
in the pipeline

1. Checkout
SSDT project
2. Build DACPAC
#2 A Pipeline Using A ‘Sidecar’ SQL Server Container
3. Spin-up container 5. Tear down
container
4. Deploy artefact
DACPAC
GitHub repository

 Using a pipeline stored under source code control
 Our first look at a declarative syntax pipeline in action
 Methods
 Jenkins built in support for powershell
 Wrapping steps in timeouts (a best practice)

Software Development In A Team
 Its highly likely that more than
one person is working on the
code at any one time
 Not a good practice for
everyone to work on the same
branch
We need a branching strategy !!!

1. Checkout SSDT
project
2. Build DACPAC
#3 A Multi-Branch Pipeline
3. Spin-up container
for the current branch
5. Tear down container
for the current branch
4. Deploy artefact
DACPAC
GitHub repository

 Our first look at a multi-branch pipeline in action
 Using ${BRANCH_NAME} to determine the
current branch
 Spinning up containers on different host ports
 The Jenkins sandbox

1. Checkout SSDT
project
2. Build DACPAC
GitHub repository
#4 A Multi-Branch Pipeline With tSQLt Testing
3. Spin-up
container
for the
current branch
6. Tear down
container
for the
current branch
4. Deploy artefact
DACPAC
5. Run tSQLt
tests

 Preparing the SQL Server container for CLR usage
 Invoking tSQLt tests
 Pipeline parameterisation
 Conditional build steps for forcing the tSQLt down
happy / un-happy test paths
 tSQLt result rendering with the Junit plugin
 “Unstable build” concept

1. Checkout SSDT
project
2. Build DACPAC
GitHub repository
#5 A Multi-Branch Pipeline With tSQLt Testing
and Parallism
3. Spin-up
container
for the
current branch
6. Tear down
container
for the
current branch
4. Deploy artefact
DACPAC
5. Run tSQLt
tests
// Parallel // Parallel
Points of
interest

The Problem With Containers
Managing state and container image size !
To minimize container image size:
 Use the smallest base images
available
 The builder pattern
To manage state:
 Use Docker volumes
The Docker “Elephant In The Room”

The Problem With Containers
Small Images Allow You To Do Some Cool Things !!!

The “Builder Pattern”
( Elephant in the room metaphor )
Source
code
Build
container
Tool chain
dacpac
Container
image
deployed
to registry
What about images that
need to be associated with
reasonably
sized data sets ?
The “Builder Pattern”

3. “Side car” Containers and Docker Volumes
1. Check out code
2. Use msbuild
to create DACPAC
3. Spin up a
container
with a volume
Name the volume
with: the branch,
build number and
project
Tests Pass
Keep the
volume
Tests fail
Remove the
volume
#6 A CI Workflow Using Docker Volumes
GitHub repository

 Docker volumes
 Jenkins nodes
 Use of the agent syntax in the declarative pipeline
 sh => for shelling out on Linux
Vs.
bat => for shelling out on Windows

Other Cool Things We Can Do
 Spin-up build slaves on a Kubernetes cluster
 Container-ize the entire build environment
 Create shared libraries to promote Groovy
script code reuse across pipelines
 Provide build status notifications via slack
 Update ticket management system (Jira)
with build status updates

etails
ChrisAdkin8
cadkin@purestorage.com
http://uk.linkedin.com/in/wollatondba

Ci with jenkins docker and mssql belgium

More Related Content

What's hot (20)

Similar to Ci with jenkins docker and mssql belgium (20)

More from Chris Adkin (17)

Recently uploaded (20)

Ci with jenkins docker and mssql belgium

Editor's Notes