Memory leak on load generator running large test

[Nova1hhh]

Brief summary

I'm observing a memory leak on a virtual machine running a k6 test. The test is launched through a script file that imports one script and uses it in scenario with ramping-arrival-rate executor. There are two stages: 1 - from zero to 17,000 rps over 2 minutes, 2 - from 17,000 to 17,000 rps for 24 hours. The script is a simple GET request to a service that responds in about 5 ms, with a check for a 200 response code.
I tried using constant-vus to achieve approximately 17k rps - the result was the same.

Using k6 0.51.0 + xk6-output-influxdb v0.4.3 + xk6-kafka v0.26.0.
Sending messages to Kafka and metrics to InfluxDB are not used.

k6 version

0.51.0

OS

Astra Linux (16 cpu\64 ram)

Docker version and image (if applicable)

No response

Steps to reproduce the problem

Run a k6 scenario that successfully generates 17k rps to the service.

Expected behaviour

Get constant resource utilization of the generator under constant load.

Actual behaviour

64 GB of memory is gradually utilized over approximately 6 hours of load at 17k rps. The response time of the targeted service, the number of requests, and resource utilization are normal.

[Nova1hhh]

additional: GET requests are sent to the same URL.

[mstoykov]

Hi @Nova1hhh, I see that you are using two different outputs, but are you also setting --no-thresholds --no-summary as recommended in the Running large tests.

You can also see #2367 and the linked issues in the first comment (among others).

[mstoykov]

I am closing this issue - I expect that the above comment by me is the case and it has been a month since the last communicaiton by @Nova1hhh