I’m using the AMI image “IE9/Chrome/Firefox - ami-0f7eae66” with the default thread setting (wpt_threads) of 1. I’d like to increase the number of threads. How do I do that without terminating and creating a new instance? Is there a config file I need to change?
Also, what’s the recommended maximum number of threads I could have running on these images? I don’t need traffic shaping.
I should add a note to the docs because I’ve largely depricated the support for running multiple threads. The Chrome/Firefox agent doesn’t support it at all and the IE agent support has a lot of caveats.
For start render detection and screen shots to work the browser window needs to be fully visible on the screen and multiple concurrent browsers breaks that.
If you REALLY want to do it you can remote into the machine and modify urlblast.ini and configure the number of threads directly. You can’t do it at all for Chrome/Firefox though.
Thanks for the reply. I’ve only just had time to get back to this.
So I can live without the screen shots. When you say render detection do you mean that WPT detects that the page has completely loaded?
I’m looking for the most cost effective way of testing hundreds of urls in the quickest time possible, without the need to spawn off multiple test agent instances.
Render detection is the start render time. You also won’t get very stable results on small EC2 instances if you run more than one at a time. Medium you might be able to go up to 2 but probably not higher. Stacking really deep requires dedicated hardware unless you don’t care about the times and are just doing some form of functional testing.
This is a good explanation of why performance testing services are so expensive because of the exact scaling problem for real browser testing: http://www.quora.com/Why-are-services-like-Watchmouse-Pingdom-and-Site24x7-so-expensive