-
Notifications
You must be signed in to change notification settings - Fork 29
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
Sparse stats graphed incorrectly? #28
Comments
Thank you for reporting, I'll investigate. |
@cwilso03 could rerun the test, save output to compressed JSON (--out=json=results.gz) and attach it please.. |
@szkiba - will do. It may take me a couple days before I'll be able to submit the additional info requested. |
@szkiba Attached are the files requested. For the dashboard command, I specified a period of 1 second ( Custom dashboard config: config.tar.gz Test results data (JSON): single-server.gz |
Sorry for the delay. I'm asking for a little more patience, I'm on vacation, and then I'll start a new job. I can probably deal with the problem in mid-August. |
No problem at all. Thanks again for the great plugin! |
Hi @cwilso03 , Thank you for your patience. The fix is ready, I have attached the report generated from your data (in the meantime, this feature has also been completed, which can be used to save the report to a file). Version v0.5.2 contains the fix. I would be happy if you have time to confirm. |
Hi @szkiba, I can confirm the graphing looks correct now. Thank you again! |
Thanks for a truly great plugin for k6! In using it this last week, I believe there may be an error in how some of the stats are being plotted. If you look at the attached screen shot, you can see there are 10 "humps" of activity in this increasing load test run. But the graphs don't show the "humps" on the same time scale. For example, in the "Generated Load" chart, the virtual user count is correct over the life of the test, ending at ~9:12 PM, but the "request rate" appears to stop at ~8:44 PM. Similarly, the three "Load Time" charts seem to show the activity "humps" compressed in the time scale.
I'm not certain of the cause, but I believe, for the Load Time stats at least (as they are my own custom metrics being gathered), it may be because the stats are not recorded on a set interval; rather, the measured time is only recorded when all of a series of queries against the server all complete successfully. Since, as the load increases the frequency of errors also increases, the later "humps" have fewer data points (as fewer of the request batches complete successfully).
Does the chart assume all data points for a given metric are occurring at the same time interval?
The text was updated successfully, but these errors were encountered: