[🐛 Bug]: High level of Workload memory usage

[EvgenyMarchuk]

What happened?

Hi all!
I have been trying to deploy selenium grid via helm chart v0.33.0 and noticed high level of workload memory usage.
In this case, azure runs autoscaling and adds a second machine to kubernates when we run tests.

I loaded selenium using kubernates helm chart by changing the following parameters:

• isolateComponents=true (for full services Router, Distributor, EventBus, SessionMap and SessionQueue, ...)
• tls
• 1 node for each browser (chrome, edge, ff)
• INFO log level
• basic authenticator true
• other params by default
• I remove the resource request and limit after deployment(browsers)

kubernates version 1.29.5
chart 0.33.0
selenium images latest

It's probably not a defect, but I need to figure out what's going wrong.
Thanks!

Command used to start Selenium Grid with Docker (or Kubernetes)

helm install selenium-grid docker-selenium/selenium-grid --set isolateComponents=true

Relevant log output

no logs

Operating System

Kebrnates(Azure)

Docker Selenium version (image tag)

latest

Selenium Grid chart version (chart version)

0.33.0

[github-actions]

@EvgenyMarchuk, thank you for creating this issue. We will troubleshoot it as soon as we can.

---

Info for maintainers

Triage this issue by using labels.

If information is missing, add a helpful comment and then I-issue-template label.

If the issue is a question, add the I-question label.

If the issue is valid but there is no time to troubleshoot it, consider adding the help wanted label.

If the issue requires changes or fixes from an external project (e.g., ChromeDriver, GeckoDriver, MSEdgeDriver, W3C), add the applicable G-* label, and it will provide the correct link and auto-close the issue.

After troubleshooting the issue, please add the R-awaiting answer label.

Thank you!

[VietND96]

Hi, thanks for your report. Can you also provide the cluster capability (CPU, memory) used?

[EvgenyMarchuk]

Standard_E8as_v4(8 cores/64Gb memory)
Standard_D8as_v4(8 cores/32Gb memory) and v5 version with 8 cores and 64Gb memory
After deploying the workload, memory usage has increased by more than 35% over 35% of availabel memory, but usage is around 10%.
You can find some information on the attached pictures. The behavior is the same for different clasters/machines.
Big eaters Distributor, Router, SessionQueue.

[VietND96]

Looks like it relates to this #2312 - Distributor is a component in part of Hub (in Hub-Node mode)
As per comment, there is a fix in Nightly build. If have time, can you refresh the setup with image tag nightly for all components and evaluate this? Then we can have fast feedback if anything can be fixed further

[EvgenyMarchuk]

If I'm not mistaken, I had the same issue
I'll check it out!

[EvgenyMarchuk]

There is some Error when I install 0.33.0 version of helm chart. Error without description

After upgrade nightly, workload memory usage is the same high