a)Look at the Hudson master logs and see if there is anything diff compared to Ubuntu
b)Check out the selenium grid logs/selenium server logs and see if you see anything

Tung

Hudson–>Ubuntu machines config is now working ,but
Hudson–>Windows machines cconfig is not working.

Let me check my environment when I am back at work on Mon.

I have made some progress with Hudson-SeleniumGrid. I did what you suggested and swapped the selenium-server-1.x.x.jar with the latest selenium-server-1.0.3-standalone.jar (with renaming after, of course) and the slave remote controls are now able to launch browser and perform selenium tests.
The problem I’m having now is that my Windows slave is unable to become a Selenium slave. Unlike the Ubuntu slaves whose “Remote FSS folder”, upon connecting, would have the following files:
1. selenium-grid
2. maven2.1-interceptor.jar
3. maven-agent.jar
4. maven-interceptor.jar
My Windows slave’s Remote FSS folder only receives the last 3:
2. maven2.1-interceptor.jar
3. maven-agent.jar
4. maven-interceptor.jar
And, as you may have guessed, the hub Hudson master does not even bother launching RC on the Windows slave (even though Hudson acknowledges the status and existence of the build executors on the Windows slave)

Do you have any suggestion on what I should try, rutvik?

Thanks,

Tung

P.S.
I recommend anyone planning to do the file swap to do it only 1 time inside the plugin “selenium.hpi”. What I did was:
- Extract the hpi plugin file, it’s actually a zip file, so any extractor will do.
- Once everything in is extracted into a folder, you need to replace selenium-server-1.x.x.jar in the following 2 locations:
- First, go to:
/WEB-INF/classes/hudson/plugins/selenium/ and look for a file called “selenium.tgz”, extract this file, go inside it, replace the selenium-server-1.x.x.jar inside “vendor” with the newer selenium-server-x.x.x.jar that you like. Now, compress it back into “tgz” file again, name it “selenium.tgz”.

- Second, go to:
/WEB-INF/lib, replace the selenium-server-1.x.x.jar inside with the newer selenium-server-x.x.x.jar that you like.

-Thirdly, compress the mothership folder back into its original zipped “.hpi” plugin format, and install the plugin as if nothing has happened.

I am not sure if it’s trivial to swap the selenium grid jar with the latest version on the Hudson plugin,but it’s worth a try since you are pretty close to your setup.Also check with the Hudson Selenium Grid plugin author if he can throw some light on it.I would also post this question on twitter.

Unfortunately both the errors that you are seeing is something that I haven’t experiencced before.

Before I go over my setup in more details, I have a question about the “Configured Environments” on the “4444″ selenium grid console page. In the standalone selenium grid installation, I would see a bunch of browsers listed. However, when the selenium grid plugin is in action, the list is empty …. I looked at the grid_configuration.yml in Hudson master’s home directory and it looks identical to the one in the standalone version. I’m not sure if this is why things are not clicking for me. …

Right now, what I have is, which Selenium grid plugin running on top of Hudson:
1. RCs registered correctly on the hub, as seen “4444 console”.
2. What I tried today is:
a) Machine [192.168.64.131] runs Hudson master and 1 Hudson dumb slave with 1 RC. (let’s call this 131)
b) Machine [192.168.64.129] runs 1 dumb slave with 1 RC (let’s call this 129)
c) Machine [192.168.64.134] runs 1 dumb slave with 1 RC (let’s call this 134)
When I tried launching a junit selenium grid test:
– If the RC on 131 (same machine as Hudson master) happened to be picked, the RC would try to launch the browser but hangs indefinitely.
– If the RC on either 129 or 134 (outside the happened to be picked), the test would crash, the RC would become “released/available” again, the Hub would log:

25-Aug-2010 1:18:16 PM com.thoughtworks.selenium.grid.HttpClient request
INFO: Remote Control replied with ‘200 / ‘Failed to start new browser session: Error while launching browser’
25-Aug-2010 1:18:16 PM com.thoughtworks.selenium.grid.hub.HubServlet forward
INFO: Responding with 200/ ‘ERROR: Could not retrieve a new session’

I’m not sure what to do about this because there isn’t enough information to find out why RCs cannot launch firefox.
Please let me know if you have any suggestions, I’m gonna try out a couple more configurations in the mean time.
Thanks,

Tung

1)Install Selenium grid and start RC on your machine(just one machine is fine) that are currently one of the slave node and see if this works correctly.(This is without Hudson into the picture).This will confirm that your Selenium Grid + RC execution works fine

2)This is slightly low possibility that something’s wrong with the selenium server version that is bundled with selenium grid.You can try to swap the selenium server jar and see if this fixes the issue (Hudson restart required)

2. [name=ll129hudson;
label=tnguyen-lucidlynx-aug-15]
A Hudson dumb-slave node running on another 10.04 machine. The node has 2 executors, the setup is quite similar to what you showed in the blog.

- When I start Hudson on apache, the Hudson master becomes the SG hub and successfully starts 2 RCs on my node, which I can verify in the http://hubURL:4444/console page:

Available Remote Controls
————————————————————————————–
Host | Port | Environment
————————————————————————————–
tnguyen-lucidlynx-aug-16.local | 48662 | /ll129hudson/tnguyen-lucidlynx-aug15/
tnguyen-lucidlynx-aug-16.local | 33924 | /ll129hudson/tnguyen-lucidlynx-aug15/
————————————————————————————–

When I run a test, the RCs do appear to move to Active RC section, however they can’t launch any browser …. The Console Output in Hudson master pauses indefinitely at:
INFO: Processing ‘1 => “*firefox”, 2 => “http://www.google.ca”, 3 => “”, cmd => “getNewBrowserSession”‘

Regards,

Tung

Regards,

Tung