-
Notifications
You must be signed in to change notification settings - Fork 506
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Bug: Artillery won't exit once the test is done #215
Comments
I've not been able to replicate with @sjb933 Is there any other output on the terminal after "All phases launched" is printed? There should be more output (either printing the number of pending request/scenarios or saying that all scenarios have completed). Does the spinner still run? Also could you try running Artillery directly just to make sure that it's Artillery that doesn't exit rather than the shell script wrapping it? |
Hey @hassy, how did you get the debug log? Is this automatically generated? I am encountering a case where the workload is done, but Artillery keeps running indefinitely. |
@gboysko debug info will be printed if you run |
We are using the latest ( |
FWIW, we are unable to reproduce the hanging condition at this time. However, we are armed with the DEBUG settings the next time it happens! |
@gboysko if you see it again please do share the logs. :-) |
I'm seeing the same behavior. The run finishes and artillery continues to run until I ctrl-c it. If I run 60 seconds at 10 virtual users artillery will generally stop. If I run 20 virtual users, it does not stop. I'm using versions: [artillery] $ artillery --version I ran with command: DEBUG=artillery:runner,runner,http,socketio artillery run add_token.json but the debug output didn't look very useful. It showed the HTTP requests and a line for each runner: runner picking scenario 0 (Add token) weight = 1 +47ms and after all the requests had finished, I get the all phases launched and ever 10 seconds the same report with NaN. \ runner All phases launched +22ms Report for the previous 10s @ 2016-12-07T01:08:11.643Z Here's my JSON: I can reproduce this consistently. Unfortunately it's on a service I'm writing for work. I have a personal/generic node/express app I use to play around to learn. I'll try and reproduce the issue on that and send you the github link. thanks, |
here's the output from [artillery] $ DEBUG=artillery:runner,runner,http,socketio artillery run add_token.json > debugoutput.txt 2>&1 |
@joeldodson very useful, thank you! I'm able to reproduce the issue with the info you provided. 👍 |
So the issue seems to be caused by a bug in Node.js v6.8.1 and above, quite possibly the bug that caused #204: Bug report: nodejs/node#9756 I haven't isolated the code in Artillery itself that triggers the bug yet, but the test case will hang reliably on Node.js 6.8.1 and above, and work fine on 6.8.0 or 4.x. @gboysko @sjb933 @joeldodson The solution is to use Node.js =<6.8.0 until the bug in Node is fixed. EDIT: Confirmed the bug in Node.js is responsible by running the test case with a Node.js binary with the patch applied. |
Great job, @hassy. I can't be certain, but I feel that I've seen the hanging condition in Node 4.x as well. However, at the time, it was also an older version of artillery. If we see the hanging condition again, we will resort to Node.js v6.8.0. |
Thanks @hassy. I'm running artillery under node 6.7 now (already had that installed) and it's working great. I'll stay with that for the time being. |
- New faster implementation of rampTo - Improved test case for rampTo (accurate expected calculation) - Add regression test for artilleryio/artillery#215 - Use arrivals@2.1.0 (fixes artilleryio/artillery#215)
I have rewritten the code inside Artillery that triggered the bug. I think we should be OK even on Node.js 6.8.1+ now. |
|
thanks @hassy! it's working great. |
Drop the latest link that will resolve the issue to help people to find it easier as this is the first link on search! |
Affected versions: tested on
1.5.0-16
and1.5.0-17
Debug log: http://hastebin.com/sasidexamo.vbs
Reported by @sjb933.
The text was updated successfully, but these errors were encountered: