When I run my test on local machine all works fine. But when I move to Jenkins I get error org.openqa.selenium.NoSuchSessionException: no such session
I do not get what can be wrong? I tested before simple test without Selenium and it works fine. In Jenkins I have linked my repo and run my test with gradle test.
Summary:
First answer is given by #ashley-frieze: You need to have Chrome Browser installed to to run test against it.
Second issue: To see the browser running, maybe take a look at Jenkins : Selenium GUI tests are not visible on Windows , where it is suggested to allow the service to "interact with desktop". Maybe that helps.
Related
I'm trying to use jenkins to create automated regression tests for a web app. Jenkins basically fetches my maven project from git and reads the pom.xml. Then the test suite is started. The problem is that my tests are using selenium web driver in combination with gekodriver in order to launch firefox and navigate on the website. But everytime I launch the tests I get those errors:
Console logs
I'm running Jenkins as a service (it's how it was launched after installation) and my tests work fine when ran through eclipse or java. So I think the problem comes from how the jenkins handles selenium webdriver.
As you can probably see, I'm on a Windows 7 professional license. Firefox and gekowebdriver are updated to the last versions.
Fixed:
Jenkins being ran as a service was the problem. In order for selenium to properly display firefox, it needed to access the desktop. Fixed the issue by simply allowing Jenkins service to use the desktop. It makes for a poor fix but it works fine.
I have an integrated selenium/sikuili automation project which I would like to run on windows agent here is the scenario: Machine (A) contains the source code Machine (B) is the agent test runner machine that has selenium grid running on it, when machine a run the test, selenium test is successfully triggered on machine B but when sikuili runs it's trying to search elements on machine(A). anyone has any idea how could I fix this issue.
thank you all in advance
You may want to use selenium grid extensions:
https://github.com/sterodium/selenium-grid-extensions
I fixed it using Jenkins, what I have basically done is creating a salve agent and point it to a remote machine(where I actually build and run my project) just made sure that I had sikuili as var in my remote machine.
friends. I try to understand how does selenium work. I written some tests with JUnit 4. These tests works fine when my app is running and I put as URL localhost:8077 where my app is started. All tests pass as I need. But now I want that my war build will depend by these tests. After these I want that the automated release (Jenkins) will also depend by selenium tests.
I don't understand some things:
Should the selenium test be included in my application or they should be assembled in a standalone application? (I read about Selenium Grid but I am not sure that I need it)?
I have one developer machine, one server with jenkins and the production server. What are the needed steps to implement Selenium test? What Url should be used in test? Is it needed to install Tomcat Server on the same server with Jenkins?
Can I invalidate the build and drop the release on Jenkins if selenium tests are not passed?
Should the selenium test be included in my application or they should be assembled in a standalone application? (I read about Selenium Grid but I am not sure that I need it)?
It depends on how you want to run your tests. Best practice is to call the test configuration inside your maven project and tests will run during your build job. You can also call your maven target after the build is completed.
Selenium Grid is useful but only if you have dedicated machines having proper setup to run those tests, else your tests may often fail. You should consider running your tests remotely on cloud using services like Saucelabs, etc.
I have one developer machine, one server with jenkins and the production server. What are the needed steps to implement Selenium test? What Url should be used in test? Is it needed to install Tomcat Server on the same server with Jenkins?
Selenium tests can be implemented with Maven on Jenkins following the step from here, here and here.
Dev or Test URLs should be used for running the test.
I think you would need Tomcat Server if you are hosting your application on the Jenkins machine or else it should not be needed.
Can I invalidate the build and drop the release on Jenkins if selenium tests are not passed?
When you include the test configuration in your project configurations, the build may fail during the build job if your test fails. You can set how you want to proceed with your build when you test fails like failing the build or pass the build with warning. You may want to get notifications for failed build also.
Ex. Jenkins marks good build as failure because of test failed
Jenkins plugins can be used based on the needs.
You may like to read more about How to do Integration tests here.
I'll do my best to answer your questions, and hopefully clear a few things up for you.
From my experience each test should be created as a new item and then configured to run after your app is deployed or periodically. No you dont need Selenium Grid. It is used for parallel execution of tests.
You'll need to configure your machines as nodes on jenkins and install the jenkins agent on the slave machines. Setting up selenium tests differs depending on how you have written the tests, for example i use maven as my build tool. So for each new test i only need to point towards svn and give the maven launch commands. The url is whatever url you use to access your app. There is a very good article on how to setup jenkins on their site, you can find it here
Yes.
I've developed some testcases localy with this setup:
Eclipse,
Selenium webdriver 2.0,
Maven 2,
TestNG,
Git
When I run the test via Maven test it works just fine. But now my next task is to move My tests to Bamboo. I've searched at Atlassians HomePage but not found any good information on how to do it.
I figure I must install webdriver on the Bamboo server? And then?
Could someone please tell me how to set up Bamboo so IT would run My tests?
Best regards, P
I'm using Ruby Selenium, so my setup is going to be different. I'll explain it anyways, in case it helps you go find your solution.
My selenium tests are written using Ruby, wrapped in RSpec (test framework). The Rspec tests are called via Rake tasks (like ANT).
On the Bamboo side, each job looks like this:
Source code checkout
Bundle install (get GEMs & dependencies)
Rake task : Bamboo Rake task that runs the named Rake task (which results in RSpec-Selenium tests running)
JUnit Parser (rake is configured to output the results of RSpec in JUnit format using CI_Reporter) parses results of tests, so Bamboo knows what passed/failed. I have this in the job config -> tasks -> final tasks section in bamboo; anything in this section will run even if a previous task failed.
apologies for the somewhat confusing title :-) Basically, I've configured my Selenium tests to run through JUnit when I build the project through maven. If I start my Selenium Server in a command prompt and then run the maven build goal in another cmd prompt then the tests runs as expected.
I figured that this process would be easy to implement in Hudson but it has proven quite problematic. I can start my selenium server as before and then configure the build to kick off via a hudson job. However, hudson doesn't seem to be picking up the selenium server - I get an error message which says the following:
Unable to obtain goal [test:test]
The reason I think that hudson isn't contacting the selenium server is because when I run the junit tests through my own command prompt with my server stopped, I get the same error message.
Its also important to note that I used to use the seleniumHQ plugin for hudson to run my tests using html files - but wish to change to JUnit as its more flexible.
Any help on this is very much appreciated!
Thanks,
Gearoid.
I recommend that you run selenium tests separately from the test:test goal.
You can create two hudson builds for your project, with the second build being triggered by the successful completion of the first build.
Why would you do this?
Unit tests complete faster, so your feedback time is mush shorter
the second build can run a shell script or ant instead of maven, allowing you easier control of your remote server
The selenium tests won't be triggered by mvn test on a developer machine, which could cause conflicts.
You can run two hudson nodes in a "farm" configuration, the second instance can run on the same server as selenium. Server affinity can be set up for specific builds.
Is there a config file with the server URL that is not being picked up running inside Hudson? Can you print the server URL to make sure it's pointing at the right place?
You could use the selenium-maven-plugin to start and stop the selenium server automatically as part of the maven build process.