Open rightchoice89 opened 7 years ago
@f2nd have you ever experienced something like this?
Hey guys,
I'll be very appreciated if you advise me smth. If you need some example artifacts I can provide them as well as a link to the test results in overload service, for compareness.
May be you used yandex.tank with jmeter by yourself and didn't faced this behaviour - in this case I will be grateful if you tell me the versions of software you had used, I will try them.
@rightchoice89 could you provide jtl file and tank log for tests with count mismatch problems? I run a lot of tests with jmeter and haven't seen such problem in 1.8.35 or 1.9.1
Hi, @f2nd Here it is - https://yadi.sk/d/P15Sw0a83N8XHY
@f2nd Any ideas / recommendations, please?
I'm trying to use yandex.tank with a jmeter as a load generator and overload service for storing test results but I encountered a strange behaviour:
I had launched a simple test and after the test completion I saw that RPS results in linux terminal are the same as in the overload. Then I took a look into a jmeter_*.jtl log that had produced during test execution in /var/loadtest/test_dir - it contained much more requests (with http code 200) then in terminal / overload.
I used different versions of yandex.tank - 1.8.35, 1.9.0, 1.9.1 and jmeter 3.0, with and without using proxy for uploading results to overload, in docker or on the server itself - the result was the same: jmeter_*.jtl log contains much more requests with successfull answers than I can see in overload.
The config I've used is the following:
Also I tried a "buffered_seconds" option - it didn't help.
Could anyone give me a piece of advice, please?
Best regards.