Private Instance No Test Results

I’ve got a private instance running and have set up a EC2 test agent using ami-561cb13e. It’s a windows 64 bit test agent. Everything is working but it’s taking 10 mins to load and then it’s saying that the test completed but there were no successful results.

So I added connectivity=LAN to the location like the documentation says and it doesn’t help. Then I saw that on the Performance Calendar site Patrick said that it could take 10 mins if the EC2 instance isn’t staying up so I added to the settings.ini:

; Use if the server will be auto-starting/stopping EC2 test agents
;ec2_key=
;ec2_secret=

; Should we automatically delete any EBS volumes marked as “available”?
; This can be used to prevent orphaned volumes but only if the account
; doesn’t expect to keep offline EBS volumes.
;ec2_prune_volumes=1

; Start an EC2 instance for every X tests in the queue (up to the location max)
EC2.ScaleFactor=100

; Default location when using EC2 auto-scaling
EC2.default=us-east-1

; Per-location minimum and maximum EC2 instance counts.
; By default the minimum will be 0 and the maximum will be 1.
; This needs to be specified for every location where you want
; different settings (EC2..min/max).
EC2.us-east-1.min=0
EC2.us-east-1.max=1
ec2_locations=0

Here’s the location file:

[locations]
1=us-east-1-group

[us-east-1-group]
1=us-east-1_wptdriver
label=“US East (N. Virginia)”
lat=38.954980
lng=-77.447956
default=us-east-1_wptdriver

[us-east-1_wptdriver]
browser=Chrome,Firefox,Safari,IE 11
label=“US East (N. Virginia)”
ami=ami-561cb13e
connectivity=LAN

I even tried a windows 32 bit system ami-c262d9ab and added this to the location file:

[locations]
1=us-east-1-group
2=virginia-group

[virginia-group]
1=virginia_wptdriver
label=“Virginia”
lat=38.954980
lng=-77.447956
default=virginia_wptdriver

[virginia_wptdriver]
browser=Chrome,Firefox,Safari
label=“Virginia”
ami=ami-c262d9ab

Here’s what the test logs are saying:

2015/02/28 05:48:57 - Test Created
2015/02/28 05:48:58 - Starting test (initiated by tester i-8623567c)
2015/02/28 05:59:58 - Test has been running for 11 minutes and it has been 100 since the last update, forcing the full test to finish.

Am I missing anything or is there any other advice anyone can give me on setting this up? Thanks for taking the time to look through it all!

What does wptdriver.ini look like?

Since I used an EC2 test agent I figured it would be already set up. I haven’t looked at it though. Would it need to be configured?

Does the EC2 machine need to have wptdriver.ini configured before it works or will it work out of the box?

To me the issue sounds like the test agent isn’t polling the server for work

If you check /install it should tell you if the agent is polling - you can also check the webserver logs for calls to getwork.php

As part of the launch configuration you’ll need to set some User Data for the AMI that are used to configure wptdriver.ini

wpt_server - the web server where WebPagetest is running (required) wpt_loc - The location name to use for wptdriver (optional - if not specified it will fall back to wpt_location or be built from the region - ec2-us-east for example) wpt_location - The location name to use for URLBlast (optional - if not specified it will be built from the region and browser - ec2-us-east-IE8 for example). wptdriver will append _wptdriver to this location ID if wpt_loc is not set. wpt_key - secret key for the specified location (optional) wpt_timeout - timeout setting for each test in seconds (optional, defaults to 60)

See bottom of here for more - Sign in - Google Accounts

/install showing that the agent is fine. The agents are updated. I have the private instance hosted on a Godaddy shared hosting if that makes a difference. How would I go about checking the webserver logs. I’ve tried to look at error logs but I’m not seeing any where they should be found.

It the web server logs you should see calls to getwork.php something like this

GET /work/getwork.php?shards=1&reboot=1&location=Local-WPTDriver&software=wpt&version$…

Also can you RDP into the agent while it’s running and watch what happens when wptdriver tries to test a page (don’t forget to restart instead of logging out otherwise you’ll get blank screenshots)

I’ll definitely give it a try. Any trick to getting into these EC2’s? What have you guys found is the best way to get in and look around?

I just SSH into the server

I am trying to setup an EC2 instance and getting the empty location issue. I do see a single one when I use ?hidden=1 at the root of the web app. I used user_data to send in my IAM user and password as well as an API_KEY that I’m setting on all the us-east-1 servers in location.ini. I copied location.ini-EC2 and only have the us-east servers enabled. I’m trying to make a very small test server just to experiment.
The behavior I see is:
[list]
[] no locations, except for using hidden flag.
[
] /install shows me several servers that display: No Agents Connected
[*] http://cl.ly/image/3x0u2r3U1F1c two examples of a test run on the same url. The older one is from command line using node.js. For example, [font=Courier]webservertest -s my.ip.xxx -k my-api-key test https://ur[/font]l and the second being one I ran through the web interface with ?hidden=1.
[/list]
I’d love some assistance and can be reached on freenode as dmarr if anyone has time to help there.

Cheers!

The EC2 machine won’t let me see the password to log into the machine. It’s a Windows 64 bit. I click get password and it says that I have to go through some crazy series of events to get the password. I really thought the EC2 machines were supposed to make it easier!

I’ve still got the problem where it’s saying it’s testing it site, and even showing the tester ID in the top left corner. But then after 10 mins it shows no results and timesout. I’ve tried everything suggested. Any chance someone could work with me more closely on this through email? I’m needing to get this set up for work as soon as possible :-/