| AUT Server can't connect to client when testexec started from jenkins [message #1794318] |
Tue, 28 August 2018 22:05  |
 Logan Winter
Messages: 6
Registered: May 2018 |
Junior Member |
|
|
Hey folks,
first of all let me say thanks for this awesome tool.
I have been using it for about 2 years now.
Something weird seems to have happened some time around the time I updated to standalone 8.6 though. Sometimes our nightly test builds will run fine a few nights in a row (at least in the sense that our application starts and the tests run :P ) but given enough runs of the job, the aut agent will get into what seems like a "zombie" state: Jenkins (who runs as a windows service that can interact with the desktop) seems to be able to connect to the agent with testexec, but after that we start to blow up and the AUT closes as soon as it opens. We were using pretty much the same configuration for the job for the past year without running into this kind of problem, so we're trying to figure out what changed, if anything, with 8.6.
The AUT agent runs on the same machine as Jenkins is hosted on, and I don't use any slaves with Jenkins. I can't use an embedded AUT agent since this is a very low memory machine + my application is a resource hog and we don't have the resources for a proper slave machine to run.
So basically, my question is, have I stumbled on some kind of bug with Jubula or have I misconfigured something?
Here is an image of my config + the logs from Jenkins when the AUT agent chokes.
You can see in the first line of the logs I am running testexec like this:
"C:\Program Files\jubula_8.6.0.262\ite\testexec.exe" -project "FOE Functional Testing" -version 1.0 -testsuite "FOE Jenkins UAT Suite" -server mktotest -port 60000 -autconfig FOE@mktotest -resultdir "C:\Program Files (x86)\Jenkins\jobs\FOE - Jubula UAT Testing\testreports" -resultname "Functional-Testing-Report" -dburl jdbc:postgresql://mktotest:5432/jubula -dbuser jubula -dbpw "password1"
|
|
|
|
| Re: AUT Server can't connect to client when testexec started from jenkins [message #1794657 is a reply to message #1794318] |
Tue, 04 September 2018 14:42  |
Logan Winter
Messages: 6
Registered: May 2018 |
Junior Member |
|
|
Thank you for the response Marvin!
We stop and restart the Agent every night using a scheduled Windows task.
I started doing that as an attempt to get around this problem; i.e. to "refresh" the agent. This didn't really fix this problem, but by making the machine jenkins is hosted on a slave of itself and using autrun, I have found an effective workaround.
That being said, I did look around in the logs you pointed out to me and found a whole bunch of stuff in the rc_swing log, that I've attached.
The rc_swing.log file is filled with errors like these, and not just from this one class,
(EDIT: "not just from this one class" meaning, it's blowing up in the same way, check AUTSwingHierarchy.checkDispatchThread, but the stack is showing me different places in the AUT start up code that seem to spawn the event, it isn't the same method in the AUT that's causing this to happen)
which implies to me that it's a sort of timing based thing or something. It looks to me (a pretty inexperienced person who wasn't around when a lot of the AUT was written ) that the AUT is initializing the Swing stuff (event dispatch thread?) before Jubula and they are fighting over that thread but I don't really know Swing or Jubula all that well.
If you can't find the cause of this, that's fine, since we have a decent enough workaround for now, but if changes are made to Jubula in response to this or if you can point to something that our program is doing, please let me know; I'm curious what's breaking :)
Thanks,
Logan
[Updated on: Tue, 04 September 2018 14:49] Report message to a moderator |
|
|
|
|
Powered by
FUDForum. Page generated in 0.02245 seconds
] 
Search
Help
Register
Login
Home
