Over past few years with Sitecore, I have worked with various technical teams to deliver Sitecore powered websites. Sometimes we all are located in the same office, but most of the times, we are located in different offices and in different countries.
Part of my job role is to make sure that all of the development work is ready for testing (QA) at agreed time intervals (end of development sprint).
Quality Assurance (QA) is an important part of delivering any project and when a QA team member discovers a defect in the system, my usual conversation with the QA team member goes something like this:
QA : “Naveed, this feature is not working on the QA site?”
Me: “Hey John (the fellow developer who did the original code), can you check why is it not working on the QA website?”
John: “I am not sure, all my code is checked-in and it works on my machine as expected!”
If you have worked on a Sitecore project, you must have had a similar conversation. The reason we have this conversation is because the deployment was done “manually”, which is an unreliable and error-prone process. A file or a package can very easily be missed during the deployment or a wrong version of a file can be deployed.
Moreover, the “manual deployment” process is very time consuming and it will increase the development cost. A typical Sitecore project may have numerous environments like QA, UAT and Production and each environment may have multiple CM and CD servers. As each server may be configured differently and may have different set of config files or config settings, the “manual” deployment process becomes even more complex. So you can see the “manual deployment” process, which started with one QA server, can quickly become very time consuming (read costly).
In an ideal world, what we would like for a Sitecore powered project, is that if someone has written a piece of code and/or updated Sitecore items, it should work on their local machine, it should work on the QA server (or any other non-production servers) and it should work on the production servers. The effort required should be minimal and nothing should be missed during deployment process. And for audit purposes every deployment should be documented.
Automated Deployments (Continuous Integration)
The best possible way to achieve this to make sure that the deployment process is fully automated from start to finish. All pieces of code and related Sitecore items should be packaged, deployed and published with “one single click” of a button.
To achieve our goal of ‘one-click-deployment’, there are tools and techniques available today (well actually since few years) which we can configure for continuous integrations and automated deployment. This will solve our 2 main issues:
- Our deployments will be reliable and consistent.
- We will be saving time and money.
If you are still doing “manual deployments” within your organization, it is time to re-think and re-evaluate this process.
The purpose of the blog series is to target newer audience/developers/tech teams, who have never done CI before and take them through a ‘hello world’ type of continuous integration process (automated builds).
For the automated deployments to work, the code udpate workflow should be as following:
1- Developers should work locally and check-in their code in the source control (use of feature branches, fix branches and pull-requests is highly recommended)
2-Continuous Integration server builds and deploys the code to the QA server
The diagram below shows a typical server architecture for the local and QA environment:
The important thing to notice here is that Sitecore items (templates, layouts, renderings, sub layouts) can be sourced control using tools such as TDS or Unicorn tools. I will be referring to TDS within the blog series.
As a minimum, automated deployment process for the above mentioned QA environment will require three servers :
1 – QA Server – Sitecore CM/CD combined on same instance
2- DB Server – Sitecore SQL server databases and MongoDB
3 – CI Server – TeamCity server for continuous build and deployment
In theory, you can use only server, but it will have performance issues for the QA website.
The tools stack that I will use to demonstrate the process in the series is as below :
- Sitecore CMS 8.2
- Visual Studios 2015
- SQL Express 2014
- Hedgehog Team Development (TDS) (licenses required)
- Sitecore Glass
- Sitecore Ship
There are other tools that can be used for the dev ops, for example, Jenkins can be used in place for TeamCity. Octopus Deploy can be used solely for deployment purposes, leaving CI server to do one job, that is to build the latest code. The decisions to choose the correct stack tools should be dependent upon your business requirements.
For simplicity purposes and for the newer audience, I will be using TeamCity for both build and deployment processes. Once you get confident with simpler process, you can enhance it/supplement it with different tools as you see fit for your project.
Assumptions for the blog series
I will assume that you starting a new project and have a blank machine/servers with only Windows on it or it is your first day at your new job. You can skip any of the below steps if you feel necessary:
- Download and install latest version of Visual Studios on your local machine or VM. At the time of writing this post, I have used VS 2015.
- Download and install latest version of SQL Express on you local machine or VM. At the time of writing this post, I have used SQL Express 2014
- Download and install TDS for VS2015 (requires commercial license, you can use 30 days evaluation for free)
I will also assume that you have RDP and FTP access to the QA, DB and CI servers as described in the architecture diagram to setup continuous integration.
As mentioned before, the blog series is intended for the ‘new audience’, developers or teams who have never done CI integrations. The series will be pretty detailed and a single blog post will not do any justice, therefore it is divided into smaller logical parts of the setup process all of which will be posted in coming weeks.
The next part of the series are:
Local Sitecore Installation
Assuming it is a new project, the first part will be to install a blank Sitecore on your local machine. There are various options for installing Sitecore on your local machine
- You can install using the official .exe installer
- You can install using the official .zip file and attaching databases
- You can install using SIM module.
- You can install using Sitecore Rocks VS extension.
Every developer will have their preferred way for installing a brand new Sitecore website, I mostly prefer official .exe installer and will be referring to this in future posts as well.
Please refer to the official documents for installing the basic Sitecore website and do not forget to do the post-deployment steps.
For the purpose of the blog series, the project will be called ‘myproject‘. Once you have installed Sitecore locally and you can navigate the link http://myproject.local on your machine you are ready for the next step. By default, the installation location will be C:\inetpub\wwwroot\myproject\
In the next part of the series we will setup a VS project locally, configure a publish profile and add XML config transforms.
- Part 1 – Continuous Integration – Why your Sitecore project deployments must be automated ?
- Part 2 – Setup and Configure Visual Studio Sitecore Project
- Part 3 – Setup and Configure TDS
- Part 4 – Setup Sitecore Glass
- Part 5 – Setup Source Control (Git)
- Part 6 – Setup QA Server, DB server and CI server
- Part 7 – Setup Continuous Integration using Team City
- Part 8 – Setup Slack Notifications with TeamCity and Bitbucket