test instances not launching

[calebmlb]

Hello,

I'm using the Amazon EC2 AMI. My tests are hanging at the front of the queue and I need some help with debugging. I have a question about something I saw in the "troubleshooting" section here:

https://sites.google.com/a/webpagetest.o...-instances

Waiting at the front of the queue
Commonly the issue is in your locations.ini. Double check that your location settings match that of which your agent is pulling the server with. Also note if you're using a key, do they match. You can check your Apache access logs for incoming requests being made by your test agent.

1. Double check that your location settings match that of which your agent is pulling the server with.
   I assume the "locations settings" are the settings found in location.ini, but how can I see what the agent is doing?
2. Also note if you're using a key, do they match.
   Which key exactly is this? The WPT API key, the EC2 secret key, the location_key in settings.ini, or something else? What two (or more) values should match?
3. You can check your Apache access logs for incoming requests being made by your test agent.
   In my access log I see periodic requests for /work/getwork.php coming from 127.0.0.1. I'm guessing these *aren't* the test agents. Does the server app poll getwork.php too?
   

Thanks!

Caleb

[pmeenan]

If you're using the EC2 server AMI then most of the troubleshooting is different since the locations.ini is already configured. There should be a /var/www/webpagetest/www/ec2/logs directory (or something similar, going from memory) where errors talking to EC2 are logged.

If I had to guess, the EC2 key and secret provided to the instance don't have permissions needed to start new instances.

[calebmlb]

Thanks for replying Patrick. That logs dir is very helpful! The problem was subnet configuration. Once I found the location-specific subnet settings I was able to launch test agents in the correct place, and all is well.

[RegusDietmar]

I have got the same problem. Can you describe where the location specific subnet config is that is blocking it - or also where EC2 key and secret provided to the instance can be configured to have permissions to allow starting instances?