Test results not posting to server

I am troubleshooting an issue on an internal WPT install and thought I would see if anyone else has experienced what I am seeing.

I have a WPT server (Version 2.5 with some customizations) which is running on a fairly standard Windows Server 2008 R2 instance on EC2. I am using AMI ami-5982d01c for my test client, spinning up with the standard set of user data.

I am kicking off tests on a schedule via the Python API for a group of about 20 sites. This runs every half hour as we are trying to collect trend data.

I’ve been running this for several months with no major issues, except that a test would occasionally complete with the message “The test completed but there were no successful results.” and a blank waterfall etc… (This was maybe 1 in 500 tests, so I didn’t worry about it).

Last week however, without any apparent change on my part, all tests are now displaying this message. I tried killing the test agent and spinning up a new one, but I am still getting the same results. I’ve verified that I can access getwork.php from the test machine, and I can see the requests being made successfully using a packet sniffer. I can also see the tests running in a browser if I RDP into the test agent box.

In the “results” folder on the server, I can see that test folders are being created and a testinfo.json.gz file and testinfo.ini file are present along with an empty “video” folder.

Based on this info, it seems like the tests are running, but somehow the results files are not posting to the server, even though connectivity doesn’t seem to be an issue.

I’ve checked permissions and even tried setting to “Full Control” for everyone on the WebPageTest folder just to eliminate that possibility.

I also tried launching tests manually via the web interface of course, but get the same exact result.

Any thoughts on other things I can try to troubleshoot this? I’ve really exhausted every possibility I can think of at this point.

Can you watch the agent while it is running a test to see if it looks like it runs correctly? Usually that happens if the agent starts crashing.

If your php.ini settings changed it could also be caused by the file size limit on POSTs being too small. Check your access log for the calls to resultImage.php and workDone.php to see if anything fishy jumps out.

What would a correct run look like? I can see the browser loading pages on the test agent, should I be seeing anything besides that?

I did a network trace and I am not seeing any posts to workdone.php from the test agent, only requests to getwork.php so I am guessing the agent is crashing for some reason before it gets to the post step.

As I mentioned, it was working great previously however, and I even tried killing the old agent instance and spinning up a new one without any luck.

My php.ini definitely hasn’t changed.

Looking at my access logs, I am seeing the expected GET and POST requests, nothing looks fishy. I tried spinning up a separate IE9 test agent as well, so I don’t think it’s agent specific at this point.

I’ve double checked php.ini and made sure my max POST and file upload thresh-holds were set high enough as well.

you can add debug=1 to urlblast.ini and see if the extra information in there is useful. You can also comment out the error_reporting(0); line in common.inc which should let any and all errors flow into the error log.

Beyond that the next step is usually to enable logging in workDone.php (debug=true at the top of the file) and take a look at the log it creates.

Thanks for the help on this. I ended up terminating the server and spinning up/configuring a fresh one, so I’m still not really sure what the issue was.