I'm experimenting with running my RF/Selenium2Library tests from Jenkins on my OSX laptop. Jenkins is installed as default as user 'jenkins'. I have the RF plugin for Jenkins installed and environment variables set, and everything works fine EXCEPT the browser never opens (neither firefox nor chrome) and the tests fail with selenium timeouts saying that such and such element never appeared.
My hunch is that I need to set a display variable somewhere to get the browser to open properly.
Here's my 'execute shell' command from Jenkins for this project:
source /Users/Shared/Jenkins/.bash_profile;
pybot -b /Users/Shared/Jenkins/robot_output/dbug.log -i SMOKE01 -d /Users/Shared/Jenkins/robot_output/ /Users/KWHome/Dev/PycharmProjects/BHI/Tests;
exit 0
Again, Selenium seems to think a browser is open somewhere and waits for the initial elements to appear, but they never do because no browser is actually open. I'm a total Jenkins newbie, so huge thanks for any advice on understanding what's going on here.
ADDED:
Here's some output from the jenkins user dbug log created after a build from Jenkins:
20140719 21:28:23.914 - INFO - Opening browser 'firefox' to base url 'https://staging.azzi-mvconnects.com/'
20140719 21:28:27.273 - DEBUG - Opened browser with session id a2414ad7-f523-7441-8a54-dd5c68f4049b
20140719 21:28:27.274 - INFO - +----- END KW: Selenium2Library.Open Browser (3360)
20140719 21:28:27.274 - INFO - +----- START KW: Selenium2Library.Maximize Browser Window [ ]
20140719 21:28:27.415 - INFO - +----- END KW: Selenium2Library.Maximize Browser Window (141)
20140719 21:28:27.416 - INFO - +----- START KW: Selenium2Library.Wait Until Element Is Visible [ ${home_login_button} ]
20140719 21:28:32.500 - INFO - +------ START KW: Selenium2Library.Capture Page Screenshot [ ]
20140719 21:28:32.909 - INFO - </td></tr><tr><td colspan="3"><img src="selenium-screenshot-1.png" width="800px">
20140719 21:28:32.909 - INFO - +------ END KW: Selenium2Library.Capture Page Screenshot (409)
20140719 21:28:32.910 - FAIL - Element 'ctl00_LoginLink' was not visible in 5 seconds
The browser appears to have 'opened', but it never appears on the desktop. I tried putting 'DISPLAY=:0' before the command to run pybot but no change.
If your Jenkins job needs to run something that displays a GUI, you cannot run that build in Jenkins which runs as a background service (whether on Windows, Mac or Linux).
(In Linux you can play tricks with Xvnc or similar fake X servers and there are even Jenkins plugins that make it simpler.)
Your alternatives are either:
Log in using the GUI session and run Jenkins in a terminal window by typing java -jar /path/to/jenkins.war. When Jenkins is started in the GUI context, any processes started by Jenkins are able to talk to the GUI system and draw windows.
Or you can set up a JNLP slave in Jenkins, then log in using the GUI session, open a web browser to access your Jenkins and start the JNLP slave that connects to Jenkins master and now that the slave is running in GUI context, you can configure the job to execute on the slave. Processes that execute in the slave will be able to talk to the GUI system and draw windows.
This question here answers my question, hands down: stackoverflow.com/questions/8787682/…. Jenkins runs the tests as a service in the background. It must do something similar on osx/unix.
We need to run headless browser. We had achieved this for Firefox. We used centOS in jenkins Slave.
I have documented it here: http://dailychitty.blogspot.in/2014/07/configuring-jenkins-agent-to-run-robot.html
Related
Summary: How do I extract Chromedriver logs when running Selenium standalone Chrome instance? I.e. interacting via Selenium API commonly on port 4444.
Details:
We are using Protractor to connect to a container running the Docker image selenium/standalone-chrome Selenium "grid". Connection info is specified via the HUB_PORT_4444_TCP_ADDR environment variable. The connection URL ends up being http://localhost:4444/wd/hub. This works fine and our tests are running successfully in Jenkins.
For completeness I'd like to extract the Chromedriver logs and attach them to the build in case we need more info for debugging test failues. How can that be done?
This question seemed like a close match but they are running Chromedriver directly. I need to ask Selenium to provide them somehow.
Log properties of standalone chrome container can be configured using JAVA_OPTS.
You can add JAVA_OPTS environment variable to standalone chrome container
name: JAVA_OPTS
value: "-Dwebdriver.chrome.logfile=<Path to log file, with file name>"
We had a shared volume mounted, and gave path to that folder for putting log file.
Used yaml file for creating container template so used in above mentioned way.
If you are using CLI to launch container, same can be passed through CLI too.
Once I've built my container with my Meteor app in it, I'd really like to be able to go
docker run me/myapp velocity test-app --ci --once --settings settings-test.json
And have it exit with 0 if successful, in which case I'll push it to docker hub, deploy it somewhere etc.
However when I try this it just hangs:
[velocity] is in continuous integration mode
[velocity] mocha is starting a mirror at http://localhost:56381/.
[velocity] *** Meteor Tools is installing ***
This takes a few minutes the first time.
[velocity] You can see the mirror logs at: tail -/app/.meteor/local/log/mocha.log
I'm using jasmine as per https://github.com/meteor-velocity/velocity-examples (I started with Mocha, but switched over to see if it made any difference).
Inspecting my .meteor/local/log files I find jasmine-client- unit.log has this at the bottom:
WARN [watcher]: [39m Pattern "/app/tests/jasmine/client/unit/**/*-+(stub|stubs|mock|mocks).+(js|coffee|litcoffee|coffee.md)" does not match any file.
WARN [karma]: [39m No captured browser, open http://localhost:9876/
INFO [karma]: [39m Karma v0.13.9 server started at http://localhost:9876/
INFO [launcher]: [39m Starting browser Chrome
ERROR [launcher]: [39m No binary for Chrome browser on your platform.
Please, set "CHROME_BIN" env variable
Parent process ( 725 ) is dead! Exiting jasmine-client-unit
Chrome clearly isn't going to be available in docker - should phantomjs be installed at this point and specified as a the running option? I would have expected this to be the case by default if the --ci option has been specified?
Thanks.
Ideally you should be using a real browser such as Chrome or Firefox to do automated testing, not PhantomJS. You can run browsers headlessly using Xvfb.
These might be useful:
http://codeutopia.net/blog/2013/07/13/headless-chromefirefox-testing-in-nodejs-with-selenium-and-xvfb/
http://elementalselenium.com/tips/38-headless
Selenium is the way to go for testing. IMO the best setup would be to use a CI server like Jenkins or TeamCity, with Xvfb installed, and have a test/deploy shell script in Jenkins such as:
#!/bin/sh
set -ex
cd $WORKSPACE
export VELOCITY_CI=1
meteor --test --settings $WORKSPACE/.deploy-staging/settings.json
cd .deploy-staging
mupx setup
mupx deploy
(Note that Xvfb is not implemented here, though). This test is not entirely working, I have yet to jump into Xvfb myself, though I know this is the right direction.
I'm using mupx to deploy my apps, which automatically creates a Docker instance on the remote server for me and handles deployment completely.
Is it possible to make Jenkins use actual browser instead of headless browser? I a running some tests written in TestNG (using Selenium webdriver). When I run the testng.xml file in Eclipse, the browser starts and the tests run. But when I use Jenkins and run the tests with maven, it doesn't start any browsers.
If your jenkins is hosted in a Windows machine, there are some special configurations you should know about services that are allow to use the interface.
By the way, the easiest way to see the browsers running is starting jenkins using the command line:
java -jar jenkins.war
In linux you could use the same command or use xvfb plugin to run browsers in background.
Hope helps
In addition to this, the main reason for not launching the browser is JNLP (java network launch protocol) , when we execute the war we can interact with the desktop applications.
Using Selenium Grid will allow you to execute the test on Jenkins but open the browser on a remote slave.
To achieve this you need to create an instance RemoteWebdriver than ChromeDriver, IEDriver etc I
For linux. If jenkins is running as a daemon, you could specify active display to connect to and run your browser on it. Check what display you could connect to:
ps e | grep -Po " DISPLAY=[\.0-9A-Za-z:]* " | sort -u
My output is:
DISPLAY=:10.0
DISPLAY=:2
DISPLAY=:2.0
Then go to your jenkins project -> Configure -> Build and add the next string above your main build configuration through "Add build step -> Execute shell"
/bin/bash -c "export DISPLAY=:10"
Edited: I've encountered the issue again recently. To resolve it:
I've given for the jenkins' user ability to interact with the desktop of my current user:
xhost +si:localuser:jenkins
so if I connect to my linux system through ssh using jenkins' user credentials, export display of my current user (export DISPLAY=:10) and run, for example google-chrome or firefox inside of putty's terminal, they are launching on my current user's desktop.
After this I've checked If I could start "mvn test" command inside of workspace/MyTests folder from my putty so it will start browser and execute tests.
At the end I've created simple script in the root of my current user:
vi ~/.startup.sh
#!/bin/bash
xhost +si:localuser:jenkins
and added it to my Xfce4 GUI: Application -> Settings -> Session and Startup -> Application Autostart, specifying command field as:
sh -c $HOME/.startup.sh
It's because of this script should work only when desktop is loaded to share my current user's display with jenkins' user. After reboot and connecting to this server through RDC desktop loads with xhost command applied. And after this jenkins could interract with desktop even when I close the RDP connection but leaving current user's session alive.
I've removed Build step in my jenkins' project configuration that was stated "Invoke top-level Maven targets". It could not start my browsers.
I've changed my "Add build step -> Execute shell" to:
export DISPLAY=:10
cd /var/lib/jenkins/workspace/MyTests
mvn test
I've tried grid also, turning selenium-server-standalone -hub and -node into daemons. But it was slower than launching browsers with WebDriver in the such way.
we are using Jenkins as CI in our project. We were running the CI from the command line using the following command
java -jar C:\\jenkins\\jenkins.war --httpPort=8085 --ajp13Port=8009
As the system needs to restart frequently, we change CI to start as a windows service.
Now we are facing the issues for Selenium test cases. Selenium test cases are not running after we make Jenkins as service. We are getting the following log and no more progress from that point
18:36:30,718 INFO [org.openqa.selenium.server.SeleniumDriverResourceHandler] Command request: getNewBrowserSession[*iexploreproxy, http://192.168.132.105:8080/, ] on session null
18:36:30,718 INFO [org.openqa.selenium.server.BrowserSessionFactory] creating new remote session
18:36:30,796 INFO [org.openqa.selenium.server.BrowserSessionFactory] Allocated session 80b95d0273ac4ea4a82860c79438f071 for http://192.168.132.105:8080/, launching...
18:36:30,796 INFO [org.openqa.selenium.server.browserlaunchers.WindowsProxyManager] Modifying registry settings...
18:36:31,781 INFO [org.openqa.selenium.server.browserlaunchers.InternetExplorerCustomProxyLauncher] Launching Internet Explorer...
Per hudson wiki, you should be running Hudson (or jenkins) as tasks rather than service for GUI testing. Check here. Look at the section GUI Testing in Windows
Following changes will resolve the problem
Update the selenium version.
Use *iexploreproxy or *piiexplore for IE instead of *iehta/*iexplore
We've been running Selenium tests via local RC for a few years, and are starting to migrate to the grid. We have been redirecting the console output of the server to our own test logs. It is quite useful for debugging. The console output looks suspiciously like Log4J - so it should be reconfigurable.
10:31:51.894 INFO - Command request: open[http://websiteundertest/IsSiteUpUp.aspx, ] on session 46c211b412474429957dbf6bebccd64d
10:31:55.323 INFO - Got result: OK on session 46c211b412474429957dbf6bebccd64d
10:31:56.330 INFO - Command request: isElementPresent[xpath=//h2, ] on session 46c211b412474429957dbf6bebccd64d
10:31:56.343 INFO - Got result: OK,true on session 46c211b412474429957dbf6bebccd64d
When re run via the hub, the console output appears only on the remote RC. The simple console redirection won't work. In scope of this question, we need to capture the remote console output someplace where the computer executing the test can get to it. We run our tests from a number of build agents via C#. We are trying grid 1 and 2. We intend to go directly to Grid 2.x.
What he meant was that this:
java -jar selenium-server-standalone-2.15.0.jar -role webdriver -hub http://127.0.0.1:4444/grid/register > mylog.txt
does not work. It will create a new file but it will be empty. I suggest using -log mylog.txt
argument when starting node or server to store log files
In the latest version of Grid, you can redirect the output when you start each of your client nodes:
java -jar selenium-server-standalone-2.15.0.jar -role webdriver -hub http://127.0.0.1:4444/grid/register > mylog.txt
This is a Windows example. You can also centralize log files on some remote server by referencing a \\serverName\mylog.txt in your redirect.
When you say:
The simple console redirection won't work...
I'm not sure I understand.