Hi,
We regularly run a suite of tests against our web applications using multiple Firefox Instances.
We are using four linux hosts, each running multiple X window sessions. In total, we are running up to 32 Firefox instances.
Occasionally we get CLIENT_GONE exceptions. I've done a number of packet captures to try and determine what the problem is and quite often I see that Firefox is downloading the Cisco OpenH264 Codec plugin that is uses to play video in the browser.
To try and resolve this problem, I have created a new profile in which the Open H264 plug-in is disabled.
However, because this appears to be a default Firefox plugin, I believe that Selenium Server overwrites the user preferences with Firefox's default preferences and the plug-in is once again enabled. I have confirmed this by seizing the browsing during the test and indeed I see the plugin has been downloaded and installed.
It's hard to pin point this as the actual cause of the problem, but it's possible to see how this could be giving rise to it. The CLIENT_GONE exceptions always correlate with a time out on page load which is recorded in the Selenium node logs. I suspect the bandwidth used to download the codec multiple times, combined with the sometimes dubious quality of our ADSL is prolonging page load times, occasionally beyond the sixty second time out.
It would be lovely if we were able to properly disable the plugin. We are considering a couple of options - one to make amend the selenium-server source code so that the codec is disabled in the copied Firefox profile - it could be enabled with a paramater? Subject to reception of the idea, it could be contributed back to "trunk". Alternatively, we could stage a local download of the Codec using amended local DNS records. Or we could use an older version of Firefox that doesn't download this Codec as a Plugin.
I'd really appreciate some feedback before we decide on how we can remedy this. It would also be great to hear from anyone experiencing similar issues.
Thank you
Daniel