Stale files under C:\Users\{my_user}\AppData\Local\Temp\ with Open Model Thread Group clutter Hard Drive

[msarlea]

Note: I've also filed this under jmeter-grpc-plugin issue 33

Script

Open Model Thread Group (bundled with JMeter 5.5) -> rate (180/min) random_arrivals(720min)
A single Unary request towards a gRPC service
SSL/TLS - enabled
Disable SSL/TLS Cert Verification - enabled
Short JSON payload
I cannot share it, because I work in regulated business environment

What happens?

I start the script in GUI mode, with low-thread-count and within a few minutes, I see that the hard drive space is going down rapidly...
Inspecting with WinDirStat, I notice there are a lot of folders/files like the ones bellow being created:

Group 1: C:\Users{my_user}\AppData\Local\Temp\protocjar{variable_long_number}\bin\protoc.exe (~7.8MB)
Group 2: C:\Users{my_user}\AppData\Local\Temp\polyglot-well-known-types{variable_long_number}\google\protobuf*.proto (~104kB)
Group 3: C:\Users{my_user}\AppData\Local\Temp\descriptor{variable_long_number}.pb.bin (~33kB)

All folders in Group 1 and Group 2 contain exactly the same files.

These folders/files are here until JMeter is closed, but, given that JMeter's way of operating the Open Model Thread Group implies opening/closing many threads, you have many stale such folders - doing 1k requests, eats up about ~8-10GB. This is not the best situation, since I'd need to soak run this for days and days at 5-10/sec rates.
I have yet to test it in a Linux environment (container as well), though.

These folders/files are also created with other types of Thread Groups, however I noticed them being stale with this one only.
Note: Issue reproduces with JMeter in NON-GUI mode

Versions

Apache JMeter 5.5
jmeter-grpc-request v1.2.5.1, but also tested with 1.2.6
OpenJDK 17 (also tested with 11)
Windows 10 v1909

[msarlea]

Issue reproduces in the same way in Linux and Docker (blazemeter/taurus image patched with this pluggin), as well.

[huynhminhtan]

This issue is stale because it has been open for 30 days with no activity.

[huynhminhtan]

This issue was closed because it has been inactive for 14 days since being marked as stale.

[arthurt-pluto]

@msarlea did you find any resolution for this? I am facing same problem

[melikhovPluto]

The same issue. Creates a lot of jars and the test is failing after some time.
Any solution?

[msarlea]

@msarlea did you find any resolution for this? I am facing same problem

Hi!
I have not found a resolution.
Moved my whole grpc testing to K6.
Learnt that in 1-2 weeks and i am good to go with Unary calls.
Hope you make it!
Cheers!