I'm trying to build a Release pipeline that is triggered by a new version of a published dotnet core tool. The trigger works fine, but I'm unable to install and run the tool in my Tasks.
CURRENTLY:
Running a Command Line Task results in a 401:
dotnet tool install -g --add-source=https://pkgs.dev.azure.com/<org-name>/_packaging/Tools/nuget/v3/index.json MyApp
C:\Program Files\dotnet\sdk\3.0.100\NuGet.targets(123,5): error : Unable to load the service index for source https://pkgs.dev.azure.com/<org-name>/_packaging/Tools/nuget/v3/index.json. [C:\Users\VssAdministrator\AppData\Local\Temp\h0g1c35v.eny\restore.csproj]
C:\Program Files\dotnet\sdk\3.0.100\NuGet.targets(123,5): error : Response status code does not indicate success: 401 (Unauthorized). [C:\Users\VssAdministrator\AppData\Local\Temp\h0g1c35v.eny\restore.csproj]
The tool package could not be restored.
Tool 'myapp' failed to install. This failure may have been caused by:
* You are attempting to install a preview release and did not use the --version option to specify the version.
* A package by this name was found, but it was not a .NET Core tool.
* The required NuGet feed cannot be accessed, perhaps because of an Internet connection problem.
* You mistyped the name of the tool.
For more reasons, including package naming enforcement, visit https://aka.ms/failure-installing-tool
This leads me to believe that I'm missing something here, making it more complicated than it needs to be, or thinking about this the wrong way. The tools in the feed can be installed locally, so I believe it's my release approach.
I'm currently looking into Personal Access Tokens (PAT)
PREVIOUSLY:
If I use the .Net Core task and the custom option:
The logs show a malformed command passed to dotnet.exe:
[command]"C:\Program Files\dotnet\dotnet.exe" "dotnet tool install -g --add-source=https://pkgs.dev.azure.com/<org-name>/_packaging/Tools/nuget/v3/index.json MyApp"
or
[command]"C:\Program Files\dotnet\dotnet.exe" "tool install -g --add-source=https://pkgs.dev.azure.com/<org-name>/_packaging/Tools/nuget/v3/index.json MyApp"
I've tried varying arguments and I tend to always see the same error message:
Could not execute because the specified command or file was not found.
Possible reasons for this include:
* You misspelled a built-in dotnet command.
* You intended to execute a .NET Core program, but dotnet-dotnet tool install -g --add-source=https://pkgs.dev.azure.com/<org-name>/_packaging/Tools/nuget/v3/index.json MyApp does not exist.
* You intended to run a global tool, but a dotnet-prefixed executable with this name could not be found on the PATH.
Your custom dotnet command is quoted and dotnet is repeated : dotnet.exe" "dotnet tool install ..." so the command is misinterpreted.
You can use the Command Line task and set the dotnet command directly :
dotnet tool install -g --add-source=https://pkgs.dev.azure.com/<org-name>/_packaging/Tools/nuget/v3/index.json MyApp
Using the .Net Core task works perfect for us.
Since the dotnet command is quoted, you need to set tool as the command and update into arguments. NuGet credentials can be provided with NuGet Auth task if needed.
Here is my walkaround:
I firstly published my private dotnet tool nupkg file also as an universal package to the same Azure Artifacts feed.
I was able to then use Universal Package task to download the .nupkg file into $(System.DefaultWorkingDirectory)/nupkgs/. This task will handle the authorization to Azure Artifacts feeds.
steps:
- task: UniversalPackages#0
displayName: 'Download mytool.cli.universal'
inputs:
downloadDirectory: '$(System.DefaultWorkingDirectory)/nupkgs/'
vstsFeed: '63d4aa2f-3ae7-4c27-8c18-aa8e3a9ff353'
vstsFeedPackage: '916d9a27-2c07-4071-8631-377f2ac08ed7'
vstsPackageVersion: 0.2.0
I then had the DotNetCoreCLI task to install my nupkg locally in agents.
steps:
- task: DotNetCoreCLI#2
displayName: 'Install mytool as a dotnet tool'
inputs:
command: custom
custom: tool
arguments: 'install --global mytool.CLI --add-source ./nupkgs --version 0.2.0'
You need to add the nuget authenticate task before you try to access the nuget feed
Related
I am trying to automate deployment of an ASP.NET WebAPI on a Linux server using the following command:
dotnet publish --configuration Release
However, when adding a new NuGet package to the solution, and then trying to run the dotnet publish command, I get an error because the compiler does not know the new package. Is there a way to tell the dotnet command to pull all NuGet packages ? (I'm kind of looking for an equivalent for pip install -r requirements.txt in python).
For information, I add the NuGet packages via VisualStudio without compiling the solution.
Edit : it seems like, unless I build the solution in VisualStudio, just adding a NuGet packet will only add the packet name and version in the file projectname.csproj.nuget.dgspec.json, but will not add the PackageReference projectname.csproj file, hince the not pulling new packets issue.
I assume you are using some CI/CD pipeline which could publish your web application somewhere.
Feels like you are missing steps before publish:
# Restore (restores nuget packages)
run: dotnet restore
# Build
run: dotnet build --configuration Release --no-restore
# Test (if you have tests in project)
run: dotnet test --no-restore --verbosity normal
# Publish
run: dotnet publish --no-restore --no-build --framework netcoreapp3.1
May be this link may be helpful: github .net CI/CD
We have Xamarin.Forms solution with iOS and UWP projects. We use Azure pipelines to build the iOS project. Until yesterday everything was working fine.
Now the build fails at the NuGet Restore step with the error:
##[error]The nuget command failed with exit code(1) and error(/Users/runner/work/1/s/"MyProjectName.UWP".csproj : error MSB4057: The target "_IsProjectRestoreSupported" does not exist in the project.
We can see that the problem occurs when trying to restore NuGet packages for the UWP project on the Mac OS build host.
Image: macOS-11
Workaround will be to exclude it from the solution, but we are using it for testing purposes and this is not a good option for us.
We had the same problem yesterday for our iOS (macos-11) and Android (macos-10.15) Pipelines.
The issue 21180 for mono seems to be the root cause, which is also referenced in a pull request to update the mono version for MacOs virtual environments. Regarding to this workflow test we gave msbuild a chance, and it works.
Solution:
Instead of NuGet restore we use directly MSBuild.
- task: MSBuild#1
inputs:
solution: 'App.sln'
configuration: 'Release'
msbuildArguments: /t:restore
Looking at the log file, MSBuild ignores the UWP project. That is the behavior that NuGet had with the older mono version 6.12.0.125. Ignoring the UWP-Project is no problem, because it can only be build on windows environments.
Project "/Users/runner/work/1/s/App.sln" on node 1 (Restore target(s)).
ValidateSolutionConfiguration:
Building solution configuration "Release|Any CPU".
/Library/Frameworks/Mono.framework/Versions/6.12.0/lib/mono/msbuild/Current/bin/NuGet.targets(315,5): warning : Skipping restore for project '/Users/runner/work/1/s/App/App.UWP/App.UWP.csproj'. The project file may be invalid or missing targets required for restore. [/Users/runner/work/1/s/App.sln]
Some additional information:
The NuGet restore task uses msbuild to find all referenced packages. If that fails, it will use the packages.config file as fallback, the pipeline step will not fail.
Instead we got an error in the build steps for Android & iOS:
error NETSDK1004: Assets file
'/Users/runner/work/1/s/../obj/project.assets.json' not found. Run a
NuGet package restore to generate this file.
Updated Answer
This can be resolved using MSBuild task instead, as the collogues mentioned below.
However, in my case this still lead to the same error. After investigating the pipeline. The root cause has been identified:
Both in MSBuild#1 and XamariniOS#2 tasks
you should target the specific iOS Project and not the solution file, like so:
"
- task: MSBuild#1
inputs:
solution: 'PathToIosProject/myproject.iOS.csproj'
configuration: '$(BuildConfiguration)'
msbuildArguments: /t:restore
" - task: XamariniOS#2
inputs:
solutionFile: PathToIosProject/myproject.iOS.csproj'
configuration: '$(BuildConfiguration)'
packageApp: true
signingIdentity: $(APPLE_CERTIFICATE_SIGNING_IDENTITY)
Old Answer
We have managed to resolve the issue. The root cause of it seems to be an update to the mono framework in the MacOS pipeline build agent.
In order to fix it, we need a script for downloading and installing an older version of the mono framework as a first step of the build pipeline like so:
link to the pipeline tasks photo
This is the code of the bash scrip used in the task:
#!/bin/bash
set -ex
MONO_MACOS_PKG_DOWNLOAD_URL='https://download.mono-project.com/archive/6.12.0/macos-10-universal/MonoFramework-MDK-6.12.0.100.macos10.xamarin.universal.pkg'
mkdir -p /tmp/mono-install
cd /tmp/mono-install
mono --version
wget -q -O ./mono-installer.pkg "$MONO_MACOS_PKG_DOWNLOAD_URL"
sudo installer -pkg ./mono-installer.pkg -target /
mono --version
ls -alh /Library/Frameworks/Mono.framework/Versions/Current
I’m using the DevOps pipeline to push my dotnet build to an Octopus cloud instance.
It worked fine until I added the dotnet SDK 3 installer to the devops pipeline
- task: UseDotNet#2
displayName: 'Install .net core 3.0.100'
inputs:
packageType: sdk
version: '3.0.100'
installationPath: $(Agent.ToolsDirectory)/dotnet
The octopus part of my azure-pipelines.yml looks like this:
- task: OctoInstaller#4
inputs:
version: 'latest'
- task: OctopusPush#4
inputs:
OctoConnectedServiceName: 'XXX.octopus.app'
Space: 'Spaces-1'
Package: '$(Build.ArtifactStagingDirectory)/$(Build.DefinitionName).$(Build.BuildNumber).zip'
Replace: 'false'
I have tried every combination of OctoInstaller that I can think of including embedded and explicit version numbers. I keep getting this error message with #4.
Failed to push package. The Octo command line tool is too old to run this task. Please use version 6.10.0 or newer, or downgrade the task to version 3.*.
With OctoInstall#3 I get another error:
Failed to push package. The process '/opt/hostedtoolcache/dotnet/dotnet' failed with exit code 150
Has anyone been able to use dotnet 3 with Octopus deploy? One option
might be to install the octo extension for dotnet but I’m not sure how
you would do that from the azure pipeline. Or to use a Windows build
instead of Ubuntu.
I also noticed an error in the build step.
The specified framework 'Microsoft.NETCore.App', version '2.0.0' was not found.
- The following frameworks were found:
3.0.0 at [/opt/hostedtoolcache/dotnet/shared/Microsoft.NETCore.App]
This means that octopus must be trying to use the dotnet 2 framework.
After looking carefully at the logs, I discovered that OctopusPush was failing because it depends on dotnet 2.0.0. I added in a yml task to install that framework and now it pushes to octopus.
It might be neater to use the dotnet octo tool, as described by TrevorBrooks however I don't know how to install dotnet extensions in the DevOps Pipeline.
Use dotnet octo https://octopus.com/blog/octopus-and-netcore
The blog article explains in detail how to make this work.
For Azure Pipelines follow this article: https://octopus.com/docs/packaging-applications/build-servers/tfs-azure-devops/using-octopus-extension
Keep in mind
The Azure DevOps extension tasks require Octo to be available on the
path when executing on a build agent and must have the .net core 2.0.0
runtime or newer installed. This may not always be possible such as
with the Azure DevOps hosted agents. In order to make this work, all
Octopus tasks will automatically attempt to download and use the
latest version of Octo tools unless they're available on the build
agent as specified above. If you would like to avoid any additional
downloads or to use a specific Octo version then you can by adding the
Octo Installer task to the start of your build definition. No attempt
will be made to download Octo if the capability is detected on your
build agent.
I'm trying to use dotnet-warp as a global tool in my .NET Core Travis-CI build, because I like the idea of a single executable so much better than a folder full of 75ish files.
I can successfully add the tool and verify there's a tools/dotnet folder in the $PATH...
But the log indicates that because .NET Core has been added recently, I'll need to restart or logout before I can actually use the tool.
Is anyone aware of a way to make this work in the Travis-CI environment?
Ran into the same issue, using the info from the Travis CI Installing Dependencies page and this comment on an issue about it, adding the following following to to my .travis.yml solved the problem:
before_script:
- export PATH=$PATH:/home/travis/.dotnet/tools
My build log:
$ export PATH=$PATH:/home/travis/.dotnet/tools
$ dotnet tool install -g dotnet-warp
You can invoke the tool using the following command: dotnet-warp
Tool 'dotnet-warp' (version '1.0.9') was successfully installed.
The command "dotnet tool install -g dotnet-warp" exited with 0.
$ cd ./src/[my project]/
The command "cd ./src/[my project]/" exited with 0.
$ dotnet-warp
Running Publish...
Running Pack...
Saved binary to "[my project]"
The command "dotnet-warp" exited with 0.
I want to create custom build task, that would do invoke dotnet core CLI tool. I've used VSTS DevOps Task SDK/node to get or install the tool:
import tl = require('vsts-task-lib/task');
async function getLibmanTool() {
let libmanExePath = tl.which('libman');
if (!libmanExePath){
console.log("Libman CLI not found. Installing..")
var dotnet = tl.tool(tl.which('dotnet', true));
await dotnet.arg(['tool', 'install', '-g', 'Microsoft.Web.LibraryManager.Cli']).exec();
}
libmanExePath = tl.which('libman', true); //this line throws, see output
return tl.tool(libmanExePath);
}
However, when I use the tool in Build Pipeline:
I'm getting following error:
Libman CLI not found. Installing..
[command]C:\hostedtoolcache\windows\dncs\2.1.105\x64\dotnet.exe tool install -g Microsoft.Web.LibraryManager.Cli
Since you just installed the .NET Core SDK, you will need to reopen the Command Prompt window before running the tool you installed.
You can invoke the tool using the following command: libman
Tool 'microsoft.web.librarymanager.cli' (version '1.0.163') was successfully installed.
##[error]Unable to locate executable file: 'libman'. Please verify either the file path exists or the file can be found within a directory specified by the PATH environment variable. Also verify the file has a valid extension for an executable file.
It look like when I install .NET Core SDK in the pipleline, it is unable to find the dotnet tool
Question:
How do I install and then safely use the dotnet core tool? Is the any way to workaround following?
Since you just installed the .NET Core SDK, you will need to reopen the Command Prompt window before running the tool you installed
There isn't any workaround to avoid reopening the CMD as far as I know.
To get it work, you could specify the install path when installing the package and then call the full path to the liman.exe. Or if you want to install it globally with "-g", then the path to the liman.exe should be "%USERPROFILE%\.dotnet\tools\liman.exe".