I'm attempting to deploy a web site to Azure using VSTS. Basically, I commit code to the GIT repo and have it setup to run CI, so it begins building as soon as I commit. However, once it hits the release section, it never copies the code to the Azure web app, rather, it gives me this line:
Info: Updating file ({projectname}\error.txt).
It doesn't copy the files I changed, but rather always just copies this file. I checked and there is indeed an error.txt file in my website directory in Azure, but it is always blank.
This build/deploy process isn't "standard" because the build step only downloads from source code, it doesn't build, because the website isn't a "web application", but rather just a "web site", meaning it doesn't need to be built.
So my build step is as follows:
Get Sources
Run on Agent - this step is empty
so the idea is that it just downloads everything from source control, that's it.
Then, my release step is as follows:
Artefacts are from build step above
deploy to environment 1 (dev)
Azure app service deploy, using "package or folder" as $(System.DefaultWorkingDirectory)/
Any idea what I might be doing wrong here?
So I actually figured this out and will leave this here in case anyone else needs it.
I admit I'm pretty new to the Azure/VSTS world, so maybe someone else is making my mistake as well.
If you don't need to "build" your project, then don't. I resolved it by simply skipping the build step altogether. What I was really after was to just download the files from source control and deploy them as-is.
In your release editor, you can specify which "artifact" you want to use to release, and one of the options is source control, which is what I did.
This would be useful for websites like mine where you don't need to build them (mine is DNN/DotNetNuke, so you don't build it before deploying).
Related
I am trying to create a release pipeline in Azure DevOps. We already have a functioning build pipeline that works well, it is able to package the build with VSBuild and publish it as an artifact. Then in the release pipeline I am using an IIS Deployment job (which includes IIS Manage and IIS Deploy tasks) and it gets that artifact to deploy.
The problem is that we already have a publish profile (.pubxml) that should take care of pretty much everything the IIS Deployment is doing (at least as far I as I understand it). So to me it seems I have two options that don't require me to refactor the project configuration itself.
I can try to mimic the settings on the IIS Deployment job to match our .pubxml as closely as possible and manually applying any changes that aren't doable through the task settings. Obviously this is not ideal as that would require us to update both when ever we make changes and it introduces a large chance of the pipeline breaking down over time.
I can scrap the idea of using IIS Deployment and just use a VSBuild task that uses arguments /p:DeployOnBuild=true /p:PublishProfile=Staging. This doesn't seem like best practices because it means my release pipeline isn't passing a build package to deploy, it is just creating a new one at each stage.
So is there a better option that would allow me to utilize the package I created with VSBuild and the .pubxml configuration together in a deploy? If that isn't possible then are either of my options the "correct" way to handle my situation or am I just missing another method of deployment I could use?
Thank you for any help or insight you can provide. Please let me know if there is any more information I can give that would be useful.
You can try using publish settings file (*.publishsettings) for your IIS deployment.
A publish settings file (.publishsettings) is different than a publishing profile (.pubxml) created in Visual Studio. A publish settings file is created by IIS or Azure App Service, or it can be manually created, and then it can be imported into Visual Studio.
To view more details, you can see:
Publish an application to IIS by importing publish settings in Visual Studio
Deploy your app to a folder, IIS, Azure, or another destination
So unfortunately there doesn't seem to be a way I can achieve everything I wanted in this. The publish profiles are required for when we build the project so without making changes to how we configure those I need to build the project whenever I want to deploy. Ultimately I went with option #2. I essentially just copied most of the build tasks used in the testing pipeline and placed those in the release pipeline with a few modified commands to actually deploy the build once finished. It all seems to work just fine but still doesn't feel like best practices. If I am missing something please let me know and I will make updates as appropriate.
I create a web project with Visual Studio.
I use GitHub desktop app to push my code to my Github
I create an Azure DevOps project throught the Azure portal to create a CI and don't change anything
When I commit a change all the CI run, the build is done with success and a new deployment is done.
To be sure everything work I did 2 differents tests
First : I created a new HTML file in my project with the other HTML files. When I commit, build and deployment run on Azure Devops without error, but I cannot see my new file in the wwwroot folder
Second : I updated a existing HTML file. When I commit, build and deployment run on Azure Devops without error, and I can see my update on the file present on the web app.
Why this difference ? I would like understood why update work correctly but not the add. I don't know where start to troubleshoot the issue
Could you please check the "build action" of the new .cshtml file?
In visual studio, right click on the file and go to properties.
Check if the build action is set to "Content", if not set it to content and commit.
Other build actions sometimes will not show any error but the view file will not be copied upon build. This is not Azure DevOps issue, rather visual studio filetype issue.
Let me know if this was the case. Good luck :)
I want to download a specific folder from my team project in VSTS and copy it to a server on premise. I've setup the vsts agent and it can copy files just fine by using the "Windows Machine File Copy", but my problem is the agent downloads my whole team project starting from the root.
In Artifacts when I choose Link an artifact source and under type choose Team Foundation Version Control, under Source (repository) I can only choose my team project $/myTeamProject in the dropdown list. I'm not able to provide a path in VSTS like $/myTeamProjet/Main/subfolder.
Is this the wrong approach? I basically want to copy some files from a subfolder in my team project in VSTS to a on premise machine, without downloading everything from the whole root folder ($/myTeamProject). It takes forever when I trigger a Release with a singe task that copies files. How can the agent map only a specific folder and not the whole root folder?
My opinion is that it's not a great approach. Your build should be publishing a set of artifacts that represents a full set of deployable bits that will remain static as you deploy them through a pipeline.
Think of this scenario: You have a release definition with a pipeline defined that goes Dev -> QA -> Prod.
You deploy to Dev. Your release definition pulls in Changeset 1234 from source control.
A few hours later, you deploy to QA. Your release definition pulls in Changeset 1234.
Someone changes some source code. You go to deploy to Prod. Your release definition pulls in Changeset 1235. Now you're deploying some stuff that hasn't been tested in a lower environment. You've just drastically increased the likelihood of a problem.
Same scenario applies if you ever want to redeploy an old version to try to repro a bug.
In short: publish that folder as an artifact as part of your build process.
In release definition, you can’t specify part of files to download from the artifacts (and the artifact source link is for you to choose artifact from which build definition).
But you can specify the files you want to copy by Windows Machine File Copy task. For the source option in Windows Machine File Copy you can specify the subfolder you want to copy, such as $(System.DefaultWorkingDirectory)/BuildDefinition/drop/Main/subfolder.
I am working on a project with multiple people, a website application which requires webpack to be built, uglified, concatenated into a few files e.g. app.min.js, style.min.css etc. - As a result of this, in an effort to prevent merge conflicts we recently added the build folder to .gitignore, under the assumption that we would be able to build during deployment.
When pushing to the Master branch, we automatically "deploy" through Semaphore CI (similar to Travis) which runs composer install, npm install, and finally "npm run build" which triggers the webpack build. This is all built and then tested on the CI side of things, and then Semaphore automatically deploys to Amazon's Elastic Beanstalk where our application is hosted.
The problem with this is, it seems Semaphore doesn't upload the build it's just tested, but rather the Master branch itself which has no built JS or CSS. I'm wondering if there's a way to push these built files to deployment as well, or if running the entire build process AGAIN on Elastic Beanstalk is the only route. It seems unnecessary to have to do that process essentially 3 times, locally, CI, and then deployment. Every time a step like this is needed on EB the actual re-instantiation time gets longer, which I'd like to keep as short as possible.
Obviously if building it a 3rd time on EB is the only way to go about this then I'll have to, just wondering if there are better solutions for this whole workflow.
I haven't worked with Semaphore CI, but you might be able to use an .ebignore file.
If you create one, the cli will use that instead of your .gitignore file.
I find in some deployment situations you want the inverse of your .gitignore (all compiled, no src). It essentially lets you pick the files from your project directory that you want to deploy, in the same way as the .gitignore file.
Edit: I just noticed the documentation on aws is lacking. It only mentions file exclusion, but you can include files too.
Edit 2: I don't think Semaphore supports the use of .ebignore, so right now this solution isn't of any use. :(
I just had a great first experience with https://deploybot.com/. The can deploy directly to elastic beanstalk. It might be interesting or you.
Trying to make my life easier, Currently we have 4 developers working in Visual Studio 2012 and we are using TFS 2012 for source control. The project we work on is a multi-tenant web application (single source directory with multiple dbs) that is a mixture of legacy, asp and vb6 com components, coupled with new C# code. We use TFS for source control and for managing User Stories and Bugs. Because of the way our site works it can not be ran or debugged locally only on the server.
Source Control is currently setup with a separate branch for each developer that's working directory is mapped to a shared network path on the dev server that has a web site pointed to it in IIS. Dev01-Dev05 etc. The developers work on projects in their branch test it using their dev website, then check in changes to their own branch and merge those into the trunk. The trunk's work space is mapped to the main dev website so that the developers can test their changes against the other customer's dev domains to test against customizations and variances in functionality based on the specific dbs the are connected to.
Very long explanation but basically each dev has a branch and a site, that are then merged into the trunk with its own site.
In order to deploy our staging server:
I compile the trunk's website via a bat file on the server
Run a windows app I built to query TFS for changesets associated with
specific WorkItems in a certain status, and copy all the files for
those changesets from the publish folder to a deployment folder.
Run another bat file on the server to use RedGate's Deployment Manager
to create a package from those new files
Go to the DM site on our network to create and deploy that release (haven't been able to get the command line tools to work for this, so I have to do it manually)
Run any SQL scripts that have been saved off in Folders that match ticket numbers on each database (10 or so customer dbs) to support the release
I have tried using TFS automated build stuff and never really got it to build the website correctly. Played around with Cruise Control also with little success. Using a mishmash of skunk works projects to do this is very time consuming and unreliable at best.
My perfect scenario would be:
Gated Checkin, Attempt build/publish every time a developer merges into the trunk, rejects and notifies developer if the build fails.
End of the day collect the TFS Items of a certain status and deploys files associated with them to the staging site
Deploy SQL scripts for those TFS items across all the customer dbs in staging
Eventually* run automated regression UI tests, create new WorkItems or emails to devs if failed
Update TFS WorkItems to new state so QA/Customers know their items are ready to test in our staging environment
Send report of what items were deployed successfully
How can I get here so that I am not spending hours preparing and deploying releases to staging and eventually production? Pretty open to potential solutions, things that would be hard to change would be the source control we are using, can't really switch to subversion or something else so we are pretty stuck with TFS.
Thanks
Went back in and started trying to get TFS to build/publish my web solution. I was able to get a build to complete successfully. adding msbuild argument /p:DeployOnBuild=True and setting the msbuild platform to x86 seemed to do the trick on that.
Then I found https://github.com/red-gate/deployment-manager-tfs which gives you a build process template to do the package and deployment using the redgate tools. After playing with that for a bit I finally got it to create, package and deploy my build to our staging environment.
Next up will be to modify the template to run some custom scripts to collect only the correct items to deploy, deploy all the sql files and then to set the workitems to the appropriate statuses after completion.
Really detailed description of your process. Thanks for sharing!
I believe you can set up TFS to have gated check-in on a single branch, which if you can setup on trunk would make sure that the merges built successfully. That could trigger msbuild, if you can get that working or a custom build job.
If you can get that working then you'd be able to use that trunk code as the artifact to send to Deployment Manager. That avoids having to assemble the files for deployment through the TFS change sets, as you'd be confident that the trunk could always build.
Are you using Deployment Manager to deploy the database from source control as well as the application?
That could be a way to further automate the process. SQL Source Control and SQL CI allow you to source control the structure of a database, keep a database up to date on each check-in, and run database unit tests. They also produce database packages for Deployment Manager, so you can deploy a release that contains both the application and the database.
If you want to send me the command you're using in step 4 to deploy the release using Deployment Manager I can help out with that. The commands I use are:
DeploymentManager.exe --create-release --server=http://localhost:81 --project="Project Name" --apiKey=XXXXXXXXXXX--version=1.1
DeploymentManager.exe --deploy-release --server=http://localhost:81 --project="Project Name" --apiKey=XXXXXXXXXXX--version=1.1 --deployto=CI-Environment-Name
That will create a release version 1.1 using the latest available packages for that project. You can optionally specify the package to be used when creating the release with
--packageversion=<package name>=<version>
--packageversion="application=1.5