JMeter tests autocancelling or delivering unexpected results

108 views
Skip to first unread message

Michael Barrett

unread,
Dec 22, 2016, 10:56:33 AM12/22/16
to RedLine13
Hi,
I'm seeing some very strange behavior in my JMeter 3.1 tests on RedLine13. More often than not, tests autocancel with the message "Test automatically cancelled due to unresponsiveness." When they do complete, it's with results that I didn't expect. For example, a recent test should have submitted 750,000 HTTPS GET requests with the following JMeter configuration:
  •    Thread Group (Threads: 25, Ramp-Up: 25, Loop Count: 300)
    • HTTP Request (Implementation: HttpClient4, Protocol: https, Method: GET)
    • CSV Data Set Config (Allow quoted data: False, Recycle on EOF: True, Stop thread on EOF: False, Sharing mode: Current thread, CSV # Lines: 100)
However, the test completed with these results:

This is on the default m3.medium server. The URLs are pointing to an API Gateway endpoint, which is using a Lambda function to fetch an S3 signed URL to an image and return a 302 response with the signed URL in its Location header. The images are small thumbnails, about 4KB each.
Locally, the JMeter tests are running fine, but we want to use RedLine13 so as not to saturate our local network.
Can someone please provide some guidance or suggestions? I've already spent over a day trying to get these tests to run.

Richard Friedman

unread,
Dec 22, 2016, 11:11:14 AM12/22/16
to Michael Barrett, RedLine13
Michael,

Can you tell me the test number or email for account you are running these on?  We can perhaps look at the logs.

Is your ramp up 25 seconds?




--
You received this message because you are subscribed to the Google Groups "RedLine13" group.
To unsubscribe from this group and stop receiving emails from it, send an email to redline13+unsubscribe@googlegroups.com.
To post to this group, send email to redl...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/redline13/17ff4d23-428c-42be-970f-713e0640db07%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

Michael Barrett

unread,
Dec 22, 2016, 11:15:49 AM12/22/16
to RedLine13, neody...@gmail.com
Hi Richard,
The test number is 35683.
The ramp-up time is 25 seconds, yes.
Thanks for looking into it!


On Thursday, December 22, 2016 at 10:11:14 AM UTC-6, Richard Friedman wrote:
Michael,

Can you tell me the test number or email for account you are running these on?  We can perhaps look at the logs.

Is your ramp up 25 seconds?



On Thu, Dec 22, 2016 at 10:56 AM, Michael Barrett <neody...@gmail.com> wrote:
Hi,
I'm seeing some very strange behavior in my JMeter 3.1 tests on RedLine13. More often than not, tests autocancel with the message "Test automatically cancelled due to unresponsiveness." When they do complete, it's with results that I didn't expect. For example, a recent test should have submitted 750,000 HTTPS GET requests with the following JMeter configuration:
  •    Thread Group (Threads: 25, Ramp-Up: 25, Loop Count: 300)
    • HTTP Request (Implementation: HttpClient4, Protocol: https, Method: GET)
    • CSV Data Set Config (Allow quoted data: False, Recycle on EOF: True, Stop thread on EOF: False, Sharing mode: Current thread, CSV # Lines: 100)
However, the test completed with these results:

This is on the default m3.medium server. The URLs are pointing to an API Gateway endpoint, which is using a Lambda function to fetch an S3 signed URL to an image and return a 302 response with the signed URL in its Location header. The images are small thumbnails, about 4KB each.
Locally, the JMeter tests are running fine, but we want to use RedLine13 so as not to saturate our local network.
Can someone please provide some guidance or suggestions? I've already spent over a day trying to get these tests to run.

--
You received this message because you are subscribed to the Google Groups "RedLine13" group.
To unsubscribe from this group and stop receiving emails from it, send an email to redline13+...@googlegroups.com.

Richard Friedman

unread,
Dec 22, 2016, 11:20:38 AM12/22/16
to Michael Barrett, RedLine13
Can you run the test with  

"Save the response output from individual tests...."  checked?

Inline image 2

That would give access to the jmeter logs and would be easier to take a look.





To unsubscribe from this group and stop receiving emails from it, send an email to redline13+unsubscribe@googlegroups.com.

To post to this group, send email to redl...@googlegroups.com.

Michael Barrett

unread,
Dec 22, 2016, 11:57:27 AM12/22/16
to RedLine13, neody...@gmail.com
I've cloned the test and run it with the "Save the response output from individual tests" box checked. Completed again, with much different results.
Test 35684

Richard Friedman

unread,
Dec 22, 2016, 1:11:37 PM12/22/16
to Michael Barrett, RedLine13
We are looking into and trying to recreate using different JMeter test. 

1 - Your test definitely fully runs, the .jtl output shows a full test run

2 - We are looking why not all your threads are reporting back  or being displayed. 

Will update as we find more.

On Thu, Dec 22, 2016 at 11:57 AM, Michael Barrett <neody...@gmail.com> wrote:
I've cloned the test and run it with the "Save the response output from individual tests" box checked. Completed again, with much different results.
Test 35684

--
You received this message because you are subscribed to the Google Groups "RedLine13" group.
To unsubscribe from this group and stop receiving emails from it, send an email to redline13+unsubscribe@googlegroups.com.
To post to this group, send email to redl...@googlegroups.com.

Richard Friedman

unread,
Dec 23, 2016, 12:52:31 AM12/23/16
to Michael Barrett, RedLine13
Michael, 

Wanted to get you an update. 

1 - We were not able to recreate the same issue, though clearly your load test is having an issue. 

2 - We did find an issue, but it was related to persisting the graph data for JMeter Threads Started graph.  We will roll up a fix for this tomorrow. 

Would be great if you could run your load test again, with that option to collect output checked.  If you can tell us when you are running it and perhaps we can connect real-time and diagnose.  Would very much like to make sure this is not an ongoing issue.   

Thanks
Rich

Michael Barrett

unread,
Dec 27, 2016, 10:06:42 AM12/27/16
to RedLine13, neody...@gmail.com
Hi Rich,
Sorry about the delay in response, was on vacation for the holidays.
I can run the test again any time today, if you're available to link up on Hangouts. Otherwise, I can run it at any time you'd like to monitor it.

On Thursday, December 22, 2016 at 11:52:31 PM UTC-6, Richard Friedman wrote:
Michael, 

Wanted to get you an update. 

1 - We were not able to recreate the same issue, though clearly your load test is having an issue. 

2 - We did find an issue, but it was related to persisting the graph data for JMeter Threads Started graph.  We will roll up a fix for this tomorrow. 

Would be great if you could run your load test again, with that option to collect output checked.  If you can tell us when you are running it and perhaps we can connect real-time and diagnose.  Would very much like to make sure this is not an ongoing issue.   

Thanks
Rich
On Thu, Dec 22, 2016 at 1:11 PM, Richard Friedman <ri...@redline13.com> wrote:
We are looking into and trying to recreate using different JMeter test. 

1 - Your test definitely fully runs, the .jtl output shows a full test run

2 - We are looking why not all your threads are reporting back  or being displayed. 

Will update as we find more.
On Thu, Dec 22, 2016 at 11:57 AM, Michael Barrett <neody...@gmail.com> wrote:
I've cloned the test and run it with the "Save the response output from individual tests" box checked. Completed again, with much different results.
Test 35684

--
You received this message because you are subscribed to the Google Groups "RedLine13" group.
To unsubscribe from this group and stop receiving emails from it, send an email to redline13+...@googlegroups.com.

To post to this group, send email to redl...@googlegroups.com.
Reply all
Reply to author
Forward
0 new messages