Question

An instance of Unicorn HTTP server runs on my server.
It loads some files in this directory :

~/apps/myapp/current/

This path is a symbolic link to the last release in this path :

~/apps/myapp/releases/234234532/

This is configured by Capistrano at deployment time.
In order to make sure that Unicorn well restarts, I do it also manually even if my deploy.rb contains a deploy:restart task called at the end since I read that there would be some issues with Capistrano against Unicorn.

My app is about to launch a Selenium instance by means of Selenium WebDriver.
However, and that’s my problem, Selenium seems to points to an old release.

To be certain, I chose to remove critical files of my process, in order to see if my app still launches well and that was the really good release I expected.
Hopefully, my apps didn’t launch and some logical errors appeared.

So now I’m certain that I point to the good release, I don’t understand the Selenium’s behaviour. I’m explaining:

The error appearing within log is :

invalid page structure: Unable to locate element: {"method":"id","selector":"sectionSearch"}
Command duration or timeout: 1.02 seconds
For documentation on this error, please visit: http://seleniumhq.org/exceptions/no_such_element.html
Build info: version: '2.21.0', revision: '16552', time: '2012-04-11 19:08:38'
System info: os.name: 'Linux', os.arch: 'amd64', os.version: '3.2.13-grsec-xxxx-grs-ipv6-64', java.version: '1.7.0_147-icedtea'
Driver info: driver.version: EventFiringWebDriver (org.openqa.selenium.NoSuchElementException)

To sum up, Selenium complains with the id element: sectionSearch as being missing in the concerned HTML page.

So, as suggests this article: http://seleniumhq.org/docs/04_webdriver_advanced.html, I increased timeout before informing that a particular html element doesn’t exist:

wait {driver.find_element(:id, 'sectionSearch')}  

For information, wait is a custom method:

def wait(timeout = 50, &block)
    Selenium::WebDriver::Wait.new(timeout: timeout).until(&block)
end

In a local computer (development environment), that works maybe because I use another SeleniumDriver: FirefoxDriver

def driver
    @driver ||= begin
      if Rails.env.production?
        driver = Selenium::WebDriver.for :remote, url: 'http://localhost:4444/wd/hub'
      else
        driver = Selenium::WebDriver.for :firefox
      end
      driver.manage.timeouts.implicit_wait = 20
      driver
    end
  end

But still the same issue with my server after even killing and restarting unicorn:

invalid page structure: Unable to locate element: {"method":"id","selector":"sectionSearch"}
    Command duration or timeout: 1.02 seconds

So, my last action was to manually change inside the server’s application current folder the line:

wait {driver.find_element(:id, 'sectionSearch')} 

by

wait {driver.find_element(:id, 'sectionSearchhhhhhhh')}

Now my expectation is to see Selenium complains with this missing element and so the error would be logical. Why doing this? Because I want to be sure that Selenium takes UPDTATED source files.

And it seems that it wasn’t the case since…same error like the preceding one appears.

Anyone having an idea?

Answer 1

This is a selenium error and you are getting it whether because your webdriver times out before reaching the element or because you have denoted the wrong selector (i.e –> you are getting error element not found). If I was you I will use xpath as the selector as it seems to be the most reliable locator out there. Let’s take for example Google’s search box as an element. You have the following html code for the box:

<input type="text" name="q" value="" id="searchText" maxlength="256"/>

The way you will form your selenium element locator using the html code above will be like this..:

driver.find_element(:xpath, "//input[@name='q']")

However this is not enough as I’ve noticed in your description above; you’ve tried setting the implicit wait. A better way of achieving the effect you want is this..

!30.times { if (driver.find_element(:xpath, "//input[@name='q']") rescue false) then break else sleep 1; end }

The code above will try 30 times, waiting 1 sec before each attempt to reach the desired element (if it fails you can capture the error by using begin/rescue clause as such:

begin
   .    
   .
 [your code]
   .
   .
!30.times { if (driver.find_element(:xpath, "//input[@name='q']") rescue false) then break else sleep 1; end }    
   .
   .
driver.find_element(:xpath, "//input[@name='q']") #so you cause an error that u can capture
   .
   .
rescue    
puts "A time-out error occurred or you have used an invalid element locator"
   .
   .
end

Hope that helps!