CI/CD using Jenkins and command line

djsunderland · January 30, 2020, 12:55pm

Hi,

I have been able to get Kantu macros running via command line locally on my Mac using Chrome. I’ve done this by simply passing the HTML file into Chrome as a file, but unfortunately, this needs to launch Chrome and there is no headless option (unlike selenium-side-runner).

My intention is to run our Selenium side files in Kantu on our EC2 instance that hosts Jenkins. I had envisaged that I would be able to install Kantu on this instance and call the HTML files via a script that is executed as part of a Jenkins job.

However, I suspect that in order to do this, I will need to install the Chrome Kantu extension on this instance (as EC2 instances do not come with a GUI access and only ssh, I will first need to install a GUI), as well as the modules.

But even if I am able to do this, as Kantu does not provide a headless mode, I am not sure if I will be able to execute these HTML files in an unattended manner. As the aim is to have a Jenkins job that when triggered will run through one or more macros that tests one of our web applications, which we can monitor via logs sent out as a post-build step.

Therefore, I would like to know how Kantu supports CI/CD through Jenkins, as I cannot see how this will work at the moment.

Many thanks for any help.

ulrich · January 31, 2020, 11:49am

Yeah, Ui.Vision needs the Chrome and/or Firefox browser to run, and the browser needs a GUI environment to work.

The reason for this is that UI.Vision supports many more commands than Selenium IDE, and some of these commands (e. g. for PDF testing) only work in a real browser, and not in a headless (limited) browser simulation.

In addition, since the the headless browser used by selenium-side-runner is not a full browser, sometimes Selenium IDE tests pass just fine when in browser mode, but fail when in headless (e. g. sometimes dropdowns have issues or select Frame fails).

But unattended execution is no problem with UI.Vision. There are two different scenarios:

You use just Selenium IDE commands => Your screen can be locked
You use UI automation features such as XClick, XMove or XType . => Your screen can not be locked. In this case you have two options:
- Use Autologin (very easy, but depending on your use case, maybe not secure enough)
- Use an unlocked virtual machine inside your locked machine. This is a very secure option. I use UI.Vision inside Ubuntu 18.04 VMs that runs on a Windows Server Hosts.

However, I suspect that in order to do this, I will need to install the Chrome Kantu extension on this instance (as EC2 instances do not come with a GUI access and only ssh, I will first need to install a GUI), as well as the modules.

Exactly… but once that is done, I can confirm that UI.Vision works fine with Jenkins