1. How can I prevent Gitlab from adding -NoProfile when creating the docker container?
2. What's the difference between using powershell with -NoProfile and without it?
3. Is there any way if I run powershell -NoProfile to somehow load/run the default profile and to revert the effect of setting -NoProfile flag?
And now the story behind these questions:
MSBuild (Visual Studio build tool command) fails to build Xamarin app when used in docker container started by Gitlab with Powershell -NoProfile.
I created a docker image for CI purspose and everything is working properly if I run the container manually, but when it is run during Gitlab runner job, it fails (more exactly msbuild /t:SignAndroidPackage fails, because of some file does not gets generated). I inspected this (put a sleep of 1 hour in gitlab-ci.yml and attached to the container on the runner machine) and found out that Gitlab starts the container with PowerShell -NoProfile...I tried that manually(start the container with -NoProfile) and I reproduced the issue.
the error is:
Could not find a part of the path 'C:\builds\gitlab\HMI.Framework\KDI\sub\HostApplicationToolkit\sub\KBle\KBle\sub\XamarinBluetooth\Source\Plugin.BLE.Android\Resources\Resource.Designer.cs'
Here the Resource.Designer.cs is missing (and it should be auto-generated during the build process)
This is the dockerFile:
# escape=`
# Use the latest Windows Server Core image with .NET Framework 4.8.
FROM mcr.microsoft.com/dotnet/framework/sdk:3.5-windowsservercore-ltsc2019
ENV VS_STUDIO_INSTALL_LINK=https://download.visualstudio.microsoft.com/download/pr/befdb1f9-8676-4693-b031-65ee44835915/c541feeaa77b97681f7693fc5bed2ff82b331b168c678af1a95bdb1138a99802/vs_Community.exe
ENV VS_INSTALLER_PATH=C:\TEMP\vs2019.exe
ENV ANDROID_COMPILE_SDK=29
# Restore the default Windows shell for correct batch processing.
SHELL ["cmd", "/S", "/C"]
ADD $VS_STUDIO_INSTALL_LINK $VS_INSTALLER_PATH
RUN %VS_INSTALLER_PATH% --quiet --wait --norestart --nocache --includeRecommended --includeOptional`
--add Microsoft.VisualStudio.Workload.NetCrossPlat `
--add Microsoft.VisualStudio.Workload.XamarinBuildTools `
|| IF "%ERRORLEVEL%"=="3010" EXIT 0
RUN del %VS_INSTALLER_PATH%
# set some util paths
RUN setx JAVA_HOME "c:\Program Files\Android\jdk\microsoft_dist_openjdk_1.8.0.25\"
RUN setx path "%path%;c:\Program Files (x86)\Android\android-sdk\tools\bin;c:\Program Files (x86)\Android\android-sdk\platform-tools"
# update android SDK with API 29
RUN echo y| sdkmanager "platforms;android-%ANDROID_COMPILE_SDK%"
#ENTRYPOINT ["powershell.exe", "-NoLogo", "-ExecutionPolicy", "Bypass"]
This is the gitlab-ci.yml
image: visualstudio2019-xamarin-ci:1.0-windowsservercore-ltsc2019
stages:
- build
- test
variables:
GIT_SUBMODULE_STRATEGY: 'recursive'
build_kdi:
stage: build
only:
- CI_CD
tags:
- docker-windows
script:
- '& "C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\MSBuild\Current\Bin\MSBuild.exe" ./src/KDI/KDI.sln /t:Restore /p:AndroidBuildApplicationPackage=true /p:Configuration=Release'
- '& "C:\Program Files (x86)\Microsoft Visual Studio\2019\Community\MSBuild\Current\Bin\MSBuild.exe" /p:Configuration=Release /t:SignAndroidPackage /property:SolutionDir="C:\builds\gitlab\HMI.Framework\KDI\src\KDI\" /p:AndroidSigningKeyStore="C:\builds\gitlab\HMI.Framework\KDI\distribution\Android\KrohneAndroidKeystore.keystore" .\src\KDI\KDI\KDI.Android\KDI.Android.csproj'
run_bb_tests:
stage: test
only:
- CI_CD
tags:
- docker-windows
script:
- docker-ci/install_and_run_bb_tests.bat "phone_ip" "5555" arg1 arg2 || true
- adb pull /storage/emulated/0/Android/data/myApp/files/Exports/BBT_Exports/BBTests_CI C:\test_results
artifacts:
paths:
- c:\test_results\BBTests_CI #expire in 1 month by default
If I start this image using : docker run -it myImage powershell everything works well, but if I run it docker run -it myImage powershell -NoProfile, msbuild fails at some step in building the xamarin app
How can I prevent Gitlab from adding -NoProfile when creating the docker container?
Apparently you can't...yet but there is a feature request in place
What's the difference between using powershell with -NoProfile and without it?
A PowerShell profile is just a file that contains things, like custom functions, for a specific user. For instance, if I were to have Function Banana(){Write-Host "Awesome"} in my PowerShell profile e.g $Home[My ]Documents\PowerShell\Profile.ps1. Whenever I open up PowerShell, it loads that function automatically. So I can open PowerShell, type banana and have the string Awesome written to stdout. If I were to start powershell with -NoProfile, and enter banana, I'd get CommandNotFoundException.
source
Is there any way if I run powershell -NoProfile to somehow load/run the default profile and to revert the effect of setting -NoProfile flag?
Maybe. And I only say maybe because I haven't personally tested it, but in theory it should work. If the profile is in the container, all you have to do is dot source it.
. "C:\path\to\powershell\profile.ps1"
dot sourcing the .ps1 file will run anything within it, inside the current scope, which includes loading functions, if any are present.
dot source docs
My expertise on Gitlab is limited, but due to my overall experience with Windows containers, PowerShell, as well as the nature of PowerShell profiles, I have doubts that the profiling is the root cause of this issue, but I don't have enough information/evidence to definitively say. Hope this helps in some way!
Related
I do have following setup:
a win PC with gitlab-runner installed (working)
a powershell script running on the same PC is starting an application
a gitlab server to connect this local PC and starting the powershell script
Now when starting the powershell script directly from the local PC, the application starts and terminates after done - working as expected. When starting the same powershell script with the gitlab server (yml-file) then I can see that the application has been started (new process in taskmanager) but it is not running as well it never terminates.
When manually end the task I see that gitlab terminates again.
Question:
what could be the root cause?
is it possible to run the powershell script with gitlab-runner? I think there is a way with the command "exec". How does the command looks like when calling the powershell script?
is it possible to run the application not in the background in order to see whats going on?
others?
thanks in advance
I think there is a bug with the gitlab runner on windows.
No matter which shell you configure in the config.toml the runner
will always use cmd.exe for an exec local run.
Specify the --shell argument to override the default cmd.exe shell:
> gitlab-runner exec shell your_job --shell pwsh
If you run this locally in your project, it outputs to .builds/, so add this to your .gitignore because git will see it and think you might want to add a submodule.
I am using cake script to build my application. While the build process all the logging informations are displayed in the console. I want to write all the console output into a specific log file in the same path where the build.ps1 is located.
Build process is like, from gitlab-ci a particular bat file will be called. That bat file will get necessary build informations and build.ps1 will be called as below.
call start /wait /i cmd /c powershell.exe -Command %PSFILE_PATH% --rebrand="app_name"
pause
[PSFILE_PATH - will have the build.ps1 file path with file name.
eg: "F:\Build\app_name\build.ps1"]
Info: I have tried using ".\build.ps1 > output.log" this works while running the build in my local machine. But, in my application build process(via gitlab-ci runner) I'm unable to use this command.
Please suggest a way(other than ".\build.ps1 > output.log") to log all the outputs printed in the console into a file while running build.ps1.
Thanks in advance.
You can persist this information from the Gitlab-ci build by saving it as an artifact. Just add this to your job's description in the gitlab-ci.yml:
artifacts:
name: 'Build output' # or whatever
when: always # or on_success/on_failure depending on your use case
paths:
- output.log
expire_in: 1 day # or 4 hours or however long you want to keep it stored
This way you can look at the output.log file via pipelines > your pipeline > artifacts.
I need to run a [Windows] Docker container as an executable, running a rather involved PowerShell script (invoking Java and .NET applications) and exiting. Docker documentation suggests using ENTRYPOINT for this purpose. So I went ahead and created a Dockerfile with the following contents:
FROM microsoft/dotnet-framework
COPY run.ps1 /
ENTRYPOINT [ "powershell.exe", "C:\\run.ps1" ]
The contents of run.ps1 (uber-simplified for this question):
gci
write-host "looks like everything is good!"
Then, I ran the following commands:
# Build the Docker image
docker build --rm -t testdockerps .
# Create/run container using above image
docker run -it testdockerps
The container ran successfully, displaying the contents of C:\ followed by the message - looks like everything is good!.
I have a couple of questions based on what my observations:
What is the default shell for a Windows based Docker container? Is there any way to set it to PowerShell, so I don't have to specify "powershell" as the first element of the ENTRYPOINT JSON array? Should I be using the SHELL command in the Dockerfile?
Creating and running the container takes about 3-4 seconds which is somewhat understandable, but after the PS1 script completes, it takes nearly a questionable 10 seconds for the container to exit and return to the command prompt. What may be the cause of this delay?
Yes you can specify powershell as default shell like below on top of DOCKERFILE
SHELL ["powershell", "-Command", "$ErrorActionPreference = 'Stop'; $ProgressPreference = 'Continue'; $verbosePreference='Continue';"]
I'm not sure you can do anything about the time it takes to spin down your VM
I am doing local Chef dev on my Windows 8.1 laptop and use VirtualBox 5.1.24 for VMs. I want to use Jenkins (v2.73.2) to run jobs that will automate the restoring of VM snapshots and running of kitchen converge, test and repeat.
I have added the VirtualBox Plugin but it doesn't let me add a node (I recall seeing somewhere that v2.73.2 of Jenkins is not supported).
I wrote some Powershell to do what I want and so after installing PowerShell Plugin created a pipeline with following PowerShell to test it:
& "C:\Program Files\Oracle\VirtualBox\vboxmanage.exe" startvm "W2008_21_06_B"
(which works in PS ISE), but fails in Jenkins job:
Building on master in workspace C:\Program Files (x86)\Jenkins\workspace\vbox
[vbox] $ powershell.exe -NonInteractive -ExecutionPolicy ByPass "& 'C:\windows\TEMP\jenkins7025720492184266246.ps1'"
vboxmanage.exe: error: Could not find a registered machine named 'W2008_21_06_B'
vboxmanage.exe: error: Details: code VBOX_E_OBJECT_NOT_FOUND (0x80bb0001), component VirtualBoxWrap, interface IVirtualBox, callee IUnknown
vboxmanage.exe: error: Context: "FindMachine(Bstr(pszVM).raw(), machine.asOutParam())" at line 573 of file VBoxManageMisc.cpp
Build step 'Windows PowerShell' marked build as failure
Finished: FAILURE
What trick am I missing please?
Regards
For the life of me, I cannot seem to get my provisioning script to execute when I run my container. Down the road, I will need to pass in arguments to the docker run command to replace 'hiiii' and '123' for multiple container deployments.
This is my docker file
FROM microsoft/aspnet:3.5-windowsservercore-10.0.14393.1198
SHELL [“powershell”, “-Command”, “$ErrorActionPreference = ‘Stop’; $ProgressPreference = ‘SilentlyContinue’;”]
COPY *.ps1 /Container/
COPY [“wwwroot”, “/inetpub/wwwroot”]
COPY [“Admin”, “/Program Files (x86)Admin”]
COPY [“Admin/web.config”, “/Program Files (x86)/Admin/web_default.config”]
#ENTRYPOINT [“powershell”]
CMD [“powershell.exe”, -NoProfile, -File, C:\Container\Start-Admin.Docker.Cmd.ps1 -Parm1 ‘Hiiii’ -parm2 ‘123’]
I have also tried the shell version of CMD as follows
CMD powershell -NoProfile -File C:\Container\Start-Admin.Docker.Cmd.ps1 -Parm1 ‘Hiiii’ -Parm2 ‘123’
This is the command I am using.
docker image build -t image:v1:v1 .
docker run --name container -p 8001:80 -d image:v1
After I create and run the container I see that the script did not run, or failed. However, I can exec into powershell on the container and run the script manually and it works fine and I see all the changes that I need.
docker exec --interactive --tty container powershell
C:\Container\Start-Admin.Docker.Cmd.ps1 -Parm1 ‘Hiiii’ -Parm2 ‘123’
I am just at a loss as to what I am missing regarding CMD.
Thanks!
I was able to get it working the way I had hoped. Though I am still working out some details in the provisioning script, this is how I ended up getting the result I wanted from the docker side of things.
FROM microsoft/aspnet:3.5-windowsservercore-10.0.14393.1198
#The shell line needs to be removed and any RUN commands need to be immediately proceeded by 'powershell' eg RUN powershell ping google.com
#SHELL [“powershell”, “-Command”, “$ErrorActionPreference = ‘Stop’; $ProgressPreference = ‘SilentlyContinue’;”]
COPY *.ps1 /Container/
COPY [“wwwroot”, “/inetpub/wwwroot”]
COPY [“Admin”, “/Program Files (x86)Admin”]
COPY [“Admin/web.config”, “/Program Files (x86)/Admin/web_default.config”]
ENV myParm1 Hiiii
ENV myParm2 123
ENTRYPOINT ["powershell", "-NoProfile", "-Command", "C:\\Container\\Start-Admin.Docker.Cmd.ps1"]
CMD ["-parm1 $Env:myParm1 -parm2 $Env:myParm2"]
The docker run command looks like this
docker run -d -p 8001:80 -e "myParm1=byeeeee" --name=container image:v1
I hope this helps someone else that is in my boat. Thanks for all the answers!
You can give this a try. ARG passes the variable to shell accessible via PS $env: variable.
ARG parmOne=Hiiii
ARG parmTwo=123
# Run a native PowerShell session and pass the script as a command.
RUN ["powershell", "-NoProfile", "-Command", "C:\\Container\\Start-Admin.Docker.Cmd.ps1 -Parm1 $env:parmOne -parm2 $env:parmTwo"]