One of the teams at Red Gate has been working very hard on a new product: Deployment Manager. Deployment Manager is a free tool that lets you deploy updates to .NET apps, services and databases through a central dashboard.
Deployment Manager has been out for a while, but I must admit that even though I work in the same building, until now I hadn’t even looked at it. My job at Red Gate is to develop and maintain some of our community sites, which involves carrying out regular deployments. One of the projects I have to deploy on a fairly regular basis requires me to send my changes to our build server, TeamCity. The output is a Zip file of the build. I then have to go and find this file, copy it across to the staging machine, extract it, and copy some of the sub-folders to other places. In order to keep track of what builds are running, I need to rename the folders accordingly.
However, even after all that, I still need to go and update the site and its applications in IIS to point at these new builds. Oh, and then, I have to repeat the process when I deploy on production. Did I mention the multiple configuration files that then need updating as well? Manually? The whole process can take well over half an hour.
I’m ready to try out a new process.
Deployment Manager is designed to massively simplify the deployment processes from what could be lots of manual copying of files, managing of configuration files, and database upgrades down to a few clicks. It’s a big promise, but I decided to try out this new tool on one of the smaller ASP.NET sites at Red Gate, Format SQL (the result of a Red Gate Down Tools week). I wanted to add some new functionality, but given it was a new site with no set way of doing things, I was reluctant to have to manually copy files around servers. I decided to use this opportunity as a chance to set the site up on Deployment Manager and check out its functionality. What follows is a guide on how to get set up with Deployment Manager, a brief overview of its features, and what I thought of the experience.
To follow along with the instructions that follow, you’ll first need to download Deployment Manager from Red Gate. It has a free ‘Starter Edition’ which allows you to create up to 5 projects and agents (machines you deploy to), so it’s really easy to get up and running with a fully-featured version.
The Initial Set Up
After installing the product and setting it up using the administration tool it provides, I launched Deployment Manager by going to the URL and port I had set it to run on. This loads up the main dashboard.
The dashboard does a good job of guiding me through the process of getting started, beginning with a prompt to create some environments.
1. Setting up Environments
The dashboard informed me that I needed to add new ‘Environments’, which are essentially ways of grouping the machines you want to deploy to. The environments that get added will show up on the main dashboard. I set up two such environments for this project: ‘staging’ and ‘live’.
2. Add Target Machines
Once I had created the environments, I was ready to add ‘target machine’s to them, which are the actual machines that the deployment will occur on.
To enable me to deploy to a new machine, I needed to download and install an Agent on it. The ‘Add target machine’ form on the ‘Environments’ page helpfully provides a link for downloading an Agent.
Once the agent has been installed, it is just a case of copying the server key to the agent, and the agent key to the server, to link them up.
3. Run Health Check
If, after adding your new target machine, the ‘Status’ flags an error, it is possible that the Agent and Server keys have not been entered correctly on both Deployment Manager and the Agent service.
You can ‘Check Health’, which will give you more information on any issues. It is probably worth running this regardless of what status the ‘Environments’ dashboard is claiming, just to be on the safe side.
4. Add Projects
Going back to the main Dashboard tab at this point, I found that it was telling me that I needed to set up a new project.
I clicked the ‘project’ link to get started, gave my new project a name and clicked ‘Create’. I was then redirected to the ‘Steps’ page for the project under the Projects tab.
5. Package Steps
The ‘Steps’ page was fairly empty when it first loaded.
Adding a ‘step’ allowed me to specify what packages I wanted to grab for the deployment. This part requires a NuGet package feed to be set up, which is where Deployment Manager will look for the packages. At Red Gate, we already have one set up, so I just needed to tell Deployment Manager about it. Don’t worry; there is a nice guide included on how to go about doing all of this on the ‘Package Feeds’ page in ‘Settings’, if you need any help with setting these bits up.
At Red Gate we use a build server, TeamCity, which is capable of publishing built projects to the NuGet feed we use. This makes the workflow for Format SQL relatively simple: when I commit a change to the project, the build server is configured to grab those changes, build the project, and spit out a new NuGet package to the Red Gate NuGet package feed. My ‘package step’, therefore, is set up to look for this package on our feed.
The final part of package step was simply specifying which machines from what environments I wanted to be able to deploy the project to.
Now the main Dashboard showed my new project and environment in a rather empty looking grid. Clicking on my project presented me with a nice little message telling me that I am now ready to create my first release!
Create a release
Next I clicked on the ‘Create release’ button in the Projects tab. If your feeds and package step(s) were set up correctly, then Deployment Manager will automatically grab the latest version of the NuGet package that you want to deploy. As you can see here, it was able to pick up the latest build for Format SQL and all I needed to do was enter a version number and description of the release.
As you can see underneath ‘Version number’, it keeps track of what version the previous release was given.
Clicking ‘Create’ created the release and redirected me to a summary of it where I could check the details before deploying.
I clicked ‘Deploy this release’ and chose the environment I wanted to deploy to and…that’s it. Deployment Manager went off and deployed it for me.
Once I clicked ‘Deploy release’, Deployment Manager started to automatically update and provide continuing feedback about the process. If any errors do arise, then I can expand the results to see where it went wrong.
That’s it, I’m done! Keep in mind, if you hit errors with the deployment itself then it is possible to view the log output to try and determine where these occurred. You can keep expanding the logs to narrow down the problem.
The screenshot below is not from my Format SQL deployment, but I thought I’d post one to demonstrate the logging output available.
One of the best bits of Deployment Manager for me is the ability to very, very easily deploy the same release to multiple machines. Deploying this same release to production was just a case of selecting the deployment and choosing the ‘live’ environment as the place to deploy to. Following on from this is the fact that, as Deployment Manager keeps track of all of your releases, it is extremely easy to roll back to a previous release if anything goes pear-shaped! You can view all your previous releases and select one to re-deploy. I needed this feature more than once when differences in my production and staging machines lead to some odd behavior.
Another option is to use the TeamCity integration available. This enables you to set Deployment Manager up so that it will automatically create releases and deploy these to an environment directly from TeamCity, meaning that you can always see the latest version up and running without having to do anything.
Machine Specific Deployments
‘What about custom configuration files?’ I hear you shout. Certainly, it was one of my concerns. Our setup on the staging machine is not in line with that on production. What this means is that, should we deploy the same configuration to both, one of them is going to break. Thankfully, it turns out that Deployment Manager can deal with this. Given I had environments ‘staging’ and ‘live’, and that staging used the project’s web.config file, while production (‘live’) required the config file to undergo some transformations, I simply added a web.live.config file in the project, so that it would be included as part of the NuGet package. In this file, I wrote the XML document transformations I needed and Deployment Manager took care of the rest. Another option is to set up ‘variables’ for your project, which allow you to specify key-value pairs for your configuration file, and which environment to apply them to. You’ll find Variables as a full left-hand submenu within the ‘Projects’ tab. These features will definitely be of interest if you have a large number of environments!
There are still many other features that I didn’t get a chance to play around with like running PowerShell scripts for more personalised deployments. Maybe next time! Also, let’s not forget that my use case in this article is a very simple one – deploying a single package. I don’t believe that all projects will be equally as simple, but I already appreciate how much easier Deployment Manager could make my life. I look forward to the possibility of moving our other sites over to Deployment Manager in the near future.
In this article I have described the steps involved in setting up and configuring an instance of Deployment Manager, creating a new automated deployment process, and using this to actually carry out a deployment. I’ve tried to mention some of the features I found particularly useful, such as error logging, easy release management allowing you to deploy the same release multiple times, and configuration file transformations.
If I had to point out one issue, then it would be that the releases are immutable, which from a development point of view makes sense. However, this causes confusion where I have to create a new release to deploy to a newly set up environment – I cannot simply deploy an old release onto a new environment, the whole release needs to be recreated.
I really liked how easy it was to get going with the product. Setting up Format SQL and making a first deployment took very little time. Especially when you compare it to how long it takes me to manually deploy the other site, as I described earlier. I liked how it let me know what I needed to do next, with little messages flagging up that I needed to ‘create environments’ or ‘add some deployment steps’ before I could continue.
I found the dashboard incredibly convenient. As the number of projects and environments increase, it might become awkward to try and search them and find out what state they are in. Instead, the dashboard handily keeps track of the latest deployments of each project and lets you know what version is running on each of the environments, and when that deployment occurred.
Finally, do you remember my complaint about having to rename folders so that I could keep track of what build they came from? This is yet another thing that Deployment Manager takes care of for you. Each release is put into its own directory, which takes the name of whatever version number that release has, though these can be customised if necessary.
If you’d like to take a look at Deployment Manager for yourself, then you can download it here.