I have a pipeline with the steps below:
Create a Resource Group
Create a Virtual Machine from an image
Copy a Python project to VM created in the previous step
This Python project is an RPA (Robotic Process Automation). Now, I need to execute this Python project inside the VM and get the outputs from it to know what is going on, what the robot is doing. This robot read some sites and internal software.
Is there a way to execute a script inside a Virtual Machine from a Pipeline and get its console outputs? Any clue?
I saw this running on Jenkins.
If you are having Microsoft host the VM for you, there may be a way to "talk" to said VM as it is hosted on Microsoft's platform. However, the only way that I know of is to deploy an agent to the VM and select it as a resource target to run tasks on, such as copying your Python project and even running it.
Have you attempted this yet?
Related
I am running a local Microsoft DevOps server along with two separate machines running test agents for nightly automated testing. I recently configured a third test agent, but have run into issues with it publishing artifacts during the Build jobs to be used by the Release jobs.
The same jobs run perfectly on the two existing test boxes, but always throw this error after the build has completed and during the artifact publishing step:
2020-09-01T21:32:29.2544245Z ##[section]Starting: Publish Artifact:
bin_drop 2020-09-01T21:32:29.2548130Z
============================================================================== 2020-09-01T21:32:29.2548196Z Task : Copy and Publish Build
Artifacts 2020-09-01T21:32:29.2548235Z Description : [DEPRECATED] Use
the Copy Files task and the Publish Build Artifacts task instead
2020-09-01T21:32:29.2548273Z Version : 1.0.32
2020-09-01T21:32:29.2548311Z Author : Microsoft Corporation
2020-09-01T21:32:29.2548365Z Help : More
Information
2020-09-01T21:32:29.2548404Z
============================================================================== 2020-09-01T21:32:30.7956739Z ***.runsettings
2020-09-01T21:32:30.8306609Z ##[error]Cannot publish artifacts from
OSX or Linux to a file share. You can change 'Artifact Type' to
'Server', use Windows, or use the 'Copy Files', 'Copy Files Over SSH',
or 'FTP Upload' task. 2020-09-01T21:32:30.8378696Z
##[section]Finishing: Publish Artifact: bin_drop
In attempting to solve this issue I did look into using the Publish Build Artifacts task instead, but its description says it would have the same problem (e.g. doesn't work on Linux etc.). To be clear, the newest test box is the same as the other two running windows 10, it is NOT running linux or OSX.
In looking at Microsoft's test agent code (I think I found the correct GitHub package here), it looks like they are using Node.js to detect the platform the agent is running on, and if it does not equal 'win32', then they are throwing this error.
So I navigated to the Node.js instance inside the agent:
\agent\externals\node\bin
\agent\externals\node10\bin
and ran this code, and verified that it did indeed return 'win32' from both Node.js instances:
const os = require('os');
console.log(os.platform());
As near as I can tell the configuration on the third box is the same as the other two Windows 10 test boxes. And this should not be throwing this error. Does anyone have an idea why the test agent might be getting a value other than 'win32' so it thinks it is not running on windows and throwing this error for me?
EDIT:
I did try removing and re-configuring the agent on the machine. On all the text boxes the agent is running in interactive mode so it can run GUI tests.
Also, I recently had to rebuild one of the other two test boxes (after I was already struggling with box 3), and that rebuild went fine. A complete re-install of everything and it is working perfectly just like it was before.
I was able to solve this issue by copying the agent from a working test machine. It seems as though the more recent agent downloaded from the DevOps server itself has a bug in it.
Copying the agent from a working machine and stripping all the _work and configuration data from it, then installing it on the new third test machine resulted in a working machine.
The test machine I rebuilt while struggling with machine three was successful because I did not do a full reformat. Just reinstalled the OS, then used the install files from its data drive to finish installing the required software on it, so I never had to download a new agent for it from the DevOps server.
I have an automation script that uses maven POM.xml to import all the dependencies needed from selenium and junit. The main test uses selenium to open a browser, verify some information, close the browser and the test ends.
When run as Junit it works fine: run as Junit test
When run as Maven Test it works fine as well: run as maven test
In both scenarios, the program opens the browser and navigates through the website as it should do for an automated test.
Now I need to integrate it to VSTS so I can visualize the overall pass/fail test on the VSTS dashboard but I'm not familiarized with this tool too much yet.
So far this is what I have managed to do:
Deploy an agent on my WindowsPC (I want to execute and deploy the project on an Azure VM or another azure instance later on) NOTE: this is the same pc I'm successfully running the program using eclipse as shown in the screenshoots above. https://learn.microsoft.com/en-us/vsts/build-release/actions/agents/v2-windows?view=vsts
Create a build definition on VSTS but when I queue the definition the build fails: build definition and the build fail.
I don't know why it can't find mt config.txt file since it is located on the same hosted agent in that same directory. I'll appreciate if someone is capable of guiding me through this process so I can run the program from the VSTS and visualize the overall tests that fail and pass on the VSTS dashboard.
UPDATE: I moved the config.txt file to the public directory and the build was successful(I still need to fix this issue because I do not want my work in a public folder).
Now the problem I have is that even though the build is successful and it looks like it is running my "3 tests", When I look at my pc, nothing is happening. it should open chrome and take a screenshot, then open Firefox and take another screenshot and finally open internet explorer and take another screenshot and save each test on different folders but it is only generating folders for chrome and internet explorer (but still those folders does not have the screenshot I'm asking, maybe because the browser is not being open on the computer.)
Here is the log: https://drive.google.com/open?id=1S_MhAUmzj8i9phPQiqS06s0_1cCRrbF0
test output report generated on my computer
test output on vsts
Look at the error message. The error message tells you precisely what the problem is: java.io.FileNotFoundException: Y:\Automation Team\CopaQA\Architecture\local\config.txt (The system cannot find the path specified)
You need to not rely on hard-coded paths.
You say you registered a build agent against your VSTS account... but did you change the agent queue for your build? If the agent queue is "Hosted", you're using Microsoft's hosted agent.
I don't know why it can't find mt config.txt file since it is located on the same hosted agent in that same directory.
It turns out that Java.IO. can't read files located on a shared network drive, I solved this by using the UNC path to that file (//"computername"/"directory"/"file.txt")
Now the problem I have is that even though the build is successful and
it looks like it is running my "3 tests", When I look at my pc,
nothing is happening.
It took me a little reading to realize that to perform UI tests my agent needs to be set up in INTERACTIVE MODE. it can be done following this guide: https://learn.microsoft.com/en-us/vsts/build-release/actions/agents/v2-windows?view=vsts
Situation in short:
Virtual Machine with Visual Studio 2013 installed. PowerShell script
runs on the VM to execute Get Latest, Build and Execute Coded UI
Tests. Windows Scheduled Task to execute PowerShell nightly.
auto-logon is enabled (or I'm doing something wrong?)
yes, I've read post Is it possible to run Coded UI tests without having to connect via remote desktop?
I've seen posts about TCM. Does this help and how can I use it in my
situation?
I made some tests in Microsoft Test Manager and I also executed and recorded them.
I've loaded these tests in a test project (and changed the script providing categories and custom checks).
I then categorized these (as Development or Acceptance).
I executed out using a PowerShell script on a VM (with Visual Studio 2013 installed) with following actions:
Get Latest
Build
Run latest build with a selected set aka category using mstest.exe
So far everything is going perfectly. All the tests pass.
However, when I create a Scheduled Task on my VM run the PowerShell script everything fails because of a missing session.
Do I have the VM (I have no knowledge of Virtual Machines) then unlock or something?
Side-Note:
I also tried to fix this with a Test Agent and Controller, but once I had installed these, all other users of TFS lost their rights, so I prefer not to do this again.
I would be very grateful if you know something that can solve this.
I spent hours on Google finding a solution for this issue, but no solution helped me.
Do I need to provide more information?
The problem you're seeming to have is that the testagent is not setup correctly. You need an active desktop session for Coded UI to be able to run (it needs it to perform all the actions such as clicks).
Microsoft has some nice info about setting up your test agent here.
But to tackle your exact problem of the test failing because of a missing session I'd suggest the following:
Run AutoLogOn.exe from the sysinternals suite (can be found at http://live.sysinternals.com/). It will automatically log in with specified user when the machine starts, and keeps the desktop session active.
Alright...I'm making progress.
I've installed test agent and controller. It's all running fine.
Next I've opened Lab Center on my own MTM to create a new environment.
Test Controller is found, but I receive the message in this post "Microsoft Test Manager cannot install test agent on these machines" when creating new Lab Center environment
File and Printer sharing exception is enabled. I don't get the other message.
I don't understand what is wrong.
Maybe I'm completely on the wrong track and it's not necessary to use the Lab Center.
Then the only remaining issue is the non-active desktop issue.
I'm doing the examples from the NServiceBus website (http://docs.particular.net/samples/step-by-step/) and when I run in Visual Studio everything works and the queues are created automatically (even if I delete them and re-run the solution than they get created automatically). The dev machine runs Windows 8.
I have an emulated Windows 2008 RC2 - when I take my solution bin folder to the server machine and try to run NServiceBus.Host.exe it eventually crashes and the reason is: "The queue does not exist or you do not have sufficient permissions to perform the operation" - even when I run with administrative privileges. When looking the the queue I see that they were not created, and if I manually create the queues than it works.
Why doesn't queues get created automatically when I run on a different machine?
Update
If I run NServiceBus.Host.exe /install than it creates the queues, and also installs it a s windows service. But what I would like to do is run it regularly. If I just run the exe, is it considered as self hosted?
You can run the NServiceBus.Host.Exe with NServiceBus.Integration argument, that should create your queues...
For more info on profiles take a look here
and here
HTH
I want to use jenkins to deploy various WARs using our single script for multiple servers.
Could you please suggest how to pass servers name to a job, so that our script can take that as an argument and start deploying on the selected server? The solution will be used to deploy the same code to 10-20 servers using our customized ant script to build these projects.
EDIT: We are using AIX servers. Want to use a drop down menu from which user can select environment IP,Port. How should I approach this?:
Maintaining txt files of environments
Using choice parameter
On selection of this env, we will use this env variable in our shell script to deploy.
To have one job start another, just use the parametrized trigger plugin. In addition, I like to run the deployment jobs on the target machine. For this I defined a slave for every target server. To be able to run a job on a specific slave and be able to choose the slave as a parameter, I use the NodeLabel Parameter Plugin.
If you want more specific tips, be more specific on what application servers you use. It would also be interesting to know if you operate under windows, linux, or other environment. The more info you give the better and more fitting the answers.