EC2 AMI agent failures on startup

I’m seeing a “Failed to launch browser” error on EC2 agents in most cases (see attachment). The server is built according to webpagetest/Server AMI.md at master · WPO-Foundation/webpagetest · GitHub. The issue appears with instances automatically created by the server and instances manually started with the appropriate user data. Some instances will execute tests correctly but the majority fail. As far as I can tell an instance never works once it displays this error.

How do I go about troubleshooting this? Is it possible to enable some type of debug logging on the agents to diagnose?

Thanks!

Can you tell what version of the agent is running? http:///getTesters.php will list the software version. For Chrome/Firefox/modern IE (wptdriver) the current version is 2.19.0.268

When the EC2 agents start up, they install the current release of Chrome and Firefox and then when they poll the server for work the server will send down a software update if they are not running the current version.

Docs for updating the server to the latest agents are here: https://sites.google.com/a/webpagetest.org/docs/private-instances#TOC-Updating-Test-Agents

Optionally, if you are running a recent release (2.19 certainly, maybe 2.18) you can enable a setting in settings.ini on the server to have it automatically update to the latest agent from the public instance as they are released: webpagetest/settings.ini.sample at master · WPO-Foundation/webpagetest · GitHub

My guess is that it is probably an older agent that has issues working with the latest release of Chrome.

Thanks for the information Patrick. Our environment is using 2.19.0.276.

After further research these issues appear to crop up when deploying agents in a private VPC subnet behind a NAT gateway. This architecture may work if the settings.ini is tweaked to hit the server over a VPC private address instead. Ultimately the NAT gateway was not required for us so I’ve simply removed it. Things seem to be working more smoothly now.