PreInfo: I have .net core web api (vs2015) mixed with just ordinary projects.
I have spent almost 2 days now to get this to work and search and tried everything I can think of, but I just cant for the live of me get the build and release in TFS online to play together.
The build (publish artifact step) says "Directory 'D:\a\1\a' is empty. Nothing will be added to build artifact 'drop'."
but the "run dot net" step says
"Published to D:\a\1\s\Operator\MobileService\root\MobileService\src\AMP.Operator.MobileService\bin\release\net452\win7-x64\publish"
...so it must be somewhere the release can pick it up but no matter what I try I can´t get it to be picked up.
Here is my build setup
dotnet run
publishing
And the realse with $(System.DefaultWorkingDirectory)/MobileService-Dev please note that I have tried every combo of $(build.artifactstagingdirectory) in the build to publish without luck but I sure this should point to the publishing folder for the build
I so hope somebody can point me to a solution. I just can´t understand how hard it is to make this work..
Within your build definition, I recommend adding a Copy Files step that will copy your the build artifacts from your msbuild results to the Build's Artifact Staging Directory before you run the Publish Artifact step.
Source Folder: $(Build.SourcesDirectory)
Contents: **\bin\$(BuildConfiguration)\**
Target Folder: $(Build.ArtifactStagingDirectory)
I am assuming that the $(BuildConfiguration) variable is custom to your definition and is probably Debug or Release. I am not sure what exactly the Run dotnet step does, but this build definition I setup published my build artifacts correctly. The Publish Build Artifacts step I'm running has the same steps as yours, except the only control option enabled is Enabled.
I am also running on TFS 2015 update 2.
You can download the files if you just want to take a look at them. Go to the build, click on the Artifacts tab, and then download as shown below:
You need to specify output argument (--output/-o) for dotnet publish command.
Arguments:
Publish -c $(BuildConfiguration) -o $(Build.ArtifactStagingDirectory)
BTW: You said “run dot net” step says “Published to D:\a\1\s....”, the files are in D:\a\1\s, no files in D:\a\1\a (one is s and another is a).
Related
The CI pipeline works well if I remove the nobuild:true option from the DotNetCoreCLI#2 task to pack the Project (ie to create a NuGet package) but I am not able to understand what special except not building the project does the nobuild option brings.
I need not want to build the Project again as the Previous task have already build the Project and locked the Assembly version of DLLs generated. I want to use the same build to create the NuGet package and to do the same I need to pass the NoBuild option but doing the same breaks the pipeline.
The pipeline gives the error that the DLLs to be packed are not present at the specified location but I tried to look at the location and I could find the DLLs.One thing that confuses me is that though I have given nobuild to be true but still the tasks shows as Building the Project.
- task: DotNetCoreCLI#2
displayName: ".NET pack"
inputs:
command: pack
packagesToPack: ${{ parameters.packagesToPack }}
nobuild: true
versioningScheme: byEnvVar
versionEnvVar: CI_Version
packDirectory: $(build.artifactStagingDirectory)\${{ parameters.packTo }}
verbosityPack: 'Normal'
Its also important to note that the same thing( nobuild:true) works on Windows Agent but it fails on Ubuntu Agent.
PS: It could be a case where windows has upgraded the agent and has caused the issue. I searched over the issue and found that one has to lock the .net SDK in the build pipeline
Thanks for the other answers that may be related to the issue but things were already taken care.
The issue was only on the Linux Environment because of an issue in .NET SDK. Refer here
The error(DLLs could not be found in the path specified ) that was being generated was correct in somehow but also it was misleading. The DLLs were being generated in Release folder at the build stage and when I was packing the DLLs they were being searched in release folder.
Though Release and release remains the same in Windows Environment but Ubuntu being case sensitive generates the Error.
The SDK implementation of .Net Core missed the IgnoreCase in the Regex option and that caused the build to break on switchin to a Linux Agent.
DotnetBuild:
Dotnet Pack:
Solution: Define the folder where to generate the DLLs in the .csproj and the automatically build and pack step would pick the DLLs from there.
For this error NU5026 ,it refers to the project being packed has not been built yet and hence cannot be packed. Please view this reference.
The file ''F:\project\bin\Debug\net461\project.exe' to be packed was not found on disk.
According to your description, you canceled the automatic build before pack. There's possibility that your build task and pack task did't run with same configuration. For example, In dotnet build task, the project is automatically built with Debug configuration, and in the pack task you set the configuration as Release.
In dotnet build task, the project is automatically built with Debug configuration.
In the dotnet pack task , the default Configuration to Package is Release
If you do not cancel the automatic build before pack, in the .net pack task the project is built in Release configuration.
So please check the log of your build task and pack task, make sure the dotnet build command and dotnet pack command use the same configuration.
I have a ClickOnce project that I'm publishing on DevOps. I've set the MSBuild Arguments property for the WinForm solution's build stage in DevOps to /target:Publish in order to trigger the creation of all the ClickOnce files:
However, that solution also contains an SSDT project, and adding the /target:Publish setting appears to then cause the build process to try publish the SSDT too. That then fails with the error:
C:\Program Files (x86)\Microsoft Visual
Studio\2017\Enterprise\MSBuild\Microsoft\VisualStudio\v15.0\SSDT\Microsoft.Data.Tools.Schema.SqlTasks.targets(1808,5):
error MSB4044: The "SqlPublishTask" task was not given a value for the
required parameter "SqlPublishProfilePath"
Presumably it's failing because there's no publish profile specified in a build parameter for the SSDT to use.
I don't want MSBuild to publish the DacPac to a server, I just want it to create the DacPac. How can I stop the /target:Publish triggering the SSDT publish, is there another build argument I can add to stop that happening?
Notes on what I tried so far to solve this, none of which has worked:
Read about the -target switch in the MS Build official docs in the MSBuild command-line reference and in the MSBuild targets section.
Looked at the code in the .csproj file to try and identify the 'Publish' sections - think Publish must also call Build.
Unticking Deploy for the SSDTs in the solution configuration in VS
Adding entries for False in the Release and Deployment configurations in the SSDT's .csproj file, and also setting that to false for the Debug configuration (as per this question)
Setting MSBuild to only publish one project using the MS Build arguments on DevOps (as per this answer)
Considered pulling the ClickOnce publish out into an entirely separate stage using Mage.exe as per this Walkthrough: Manually deploy a ClickOnce application
Tried to create a publish profile that doesn't actually publish, so that the publish stage can complete (was looking at this question for ideas on that and also the official documentation for SqlPackage.exe)
Eventually I solved the issue above a completely different way. Instead of getting MSBuild to do what I wanted it to, I instead split the solution configuration in two, with one stage for the databases and one for the WinForms project without the databases.
I then used two separate VS Build stages on DevOps with only the WinForms stage still having /target:Publish set.
I've written that up here, but would still like to know the answer to whether it's possible to tell MSBuild not to build the SSDTs when the target is set to Publish?
We have started using Appveyor for CI with our Github repo and all has went fine. We have our build script working (appveyor.yml) in the repo, and it successfully builds and produces an artifact. Now on to my question
We are trying to pair our builds now with codecov.io which says it will scan it during the CI build. And it does support appveyor, it merely says to add this to the yml
after_test:
- bash <(curl -s https://codecov.io/bash)
Which we did, but nothing happens. Does this mean i need an actual test script to run, before it will send off the code to codecov? It is a C# project, and we have it compile, and generate the .exe which we then package into a zip which can be downloaded.
What are the point of these "test scripts" when we already know it compiles and produces an exe? I am a bit confused on how to use this properly..
I believe here https://github.com/codecov/example-csharp is good example of running codecov tests on C# project with working Appveyor configuration. Please look at readme.md file in that repo for greater detail.
From the Documentation
You need to add the OpenCover nuget package to your Visual Studio solution which is used to generate code coverage analysis:
PM> Install-Package OpenCover
Secondly, you need to either write a PowerShell script (if you intend to generate code coverage and upload the result interactively) or you need to add a few entries in you CI config file (if you intend to let your CI generate the coverage).
I have been playing around with TeamCity to get a CI environment up and running.
I started by following Troy Hunt's 'You're deploying wrong', which was very useful, however I wanted to split the packaging and deployment into 2 seperate steps, for the following reasons:
I wanted to pass some additional flags to msdeploy, which isnt possible (to m,y knowledge) by using the MSBuild Package and Deploy that Troy describes.
I can easily disable the Second Build Step i.e. deployment, if I ever need to build the package but not deploy it.
I wanted to use the -skip flag on msdeploy to prevent it from deleting certain folders, which again I couldn't find any method of doing without passing as an argument to the command line.
So, in my first MSBuild step I just have the parameters:
/P:Configuration=%env.Configuration%
/P:VisualStudioVersion=11.0
/P:IgnoreDeployManagedRuntimeVersion=True
And then I have a second Build Step that uses a command line build runner to execute the following msdeploy command:
"C:\Program Files\IIS\Microsoft Web Deploy V3\msdeploy.exe" -verb:sync -source:package="C:\ProgramData\JetBrains\TeamCity\system\artifacts\MyProject\%system.teamcity.buildConfName%\%teamcity.build.id%\MyProject.Web.csproj.zip" -dest:auto,ComputerName='https://devserver:8172/msdeploy.axd?site=MyWebsite',UserName='domain\username',Password='password',IncludeAcls='False',AuthType='Basic' -skip:objectName=dirPath,absolutePath=media$ -disableLink:AppPoolExtension -disableLink:ContentExtension -disableLink:CertificateExtension -retryAttempts=2 -allowuntrusted
The problem with this is that apparently TeamCity doesnt publish the artifacts until all the build steps are complete, so therefore the Command Line process fails because the package zip file doesnt actually exist at that point.
I have read something about publishing artifacts whilst the Build is still in progress but that does seem like a bit of a hack.
Any advice would be greatly appreciated.
You would be better have two builds vs two build steps.
The first (Build A) would have 1 build step that would do the building then the second (Build B) would just do the deployment in one build step using the artifacts from the first.
So under the first configuration tab of build A, you would specify the artifacts that would need to be made available from the first build. You can then run the build and confirm that under the artifacts section everything you need is available. (These will show when build A has completed running).
Then under the dependencies section (Can't remember the exact name, and I'm away from my TC instance) of Build B you can set it up to use the artifacts of build A which would then be used for the deployment.
Once you have all that working, you could add a build trigger to have Build B run after a successfull execution of Build A, then if you at one point just wanted to run Build A, disable the trigger on build B or pause the Build B configuration which will stop the trigger from firing.
My setup in TeamCity:
I've got a "CI configuration" running a singular build step that uses a MSBuild script checked-in to the from the project. The configuration has a number of sub targets such as unit tests, deployment to test site etc.
I also have a "Deploy configuration" with a singular target using another MSBuild script that only does deployment (using msdeploy). This is triggered manually.
My problem:
The "Deploy configuration" need to deploy a revision based on the latest pinned build from "CI configuration". The whole idea is essentially to get a one-click-deployment functionality based on latest pinned build in CI. Does anyone know how to do this in MSBuild using TeamCity?
Im guessing I need to use "Snapshot Dependencies" to get the latest pinned revision, but I have not been able to figure out how to deploy that snapshop in the MSBuild script.
Also I'd like to specify that I would like to do this within MSBuild script and not by using the "Custom build" dialog in TeamCity.
Ok, figured it out on my own.
You need to set a row in the "Artifacts path" like so "*/ => source" in the CI configuration.
Then add a "Artifact dependency" in the Deploy configuration pointing to the CI configuration in the "Depend on" select and set "Get artifacts from" to "Last pinned build" and then point to the "Artifact rules" like so: "source => source".
And at last, inside my MSBuild file, I added "/source/", to my project path in the Target-node with the deploy.
Works like a sharm every deployment. :-)