Closed chandupranayp closed 1 month ago
@chandupranayp, thank you for creating this issue. We will troubleshoot it as soon as we can.
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!
Hi @chandupranayp, autoscaling is on top of your own existing KEDA? If yes, which KEDA version did you use?
Hi @vietnd96, yes, autoscaling is on top of existing KEDA and KEDA version we are using is 2.9.3
Yes, so I suggest that you should upgrade KEDA to recent version, now is 2.13.0
to test and confirm
If take a look at KEDA changelog https://github.com/kedacore/keda/blob/main/CHANGELOG.md - between 2.9.3 - 2.13.0 there are few fixes for Selenium Grid Scaler
@vietnd96, Thanks for your quick feedback.
I need to wait for a couple more weeks to update the KEDA version due to some other dependencies. I can only test and confirm after that. However, after reviewing the changelog you provided, I didn't find any fixes related to the EDGE issue I am currently facing. Do you have any suggestions for other possible issues that I can try to fix and test before proceeding with the KEDA upgrade?
Ah yes, as you mentioned after upgrade 4.16.1. In this version, in chart 0.26.3 there was a change that updated default value autoscaling.scalingStrategy.strategy
from accurate
to default
In case you are using scalingType: job
and facing this issue, can you try to change it back accurate
Noted: in the latest chart 0.27.0, this default value changed back accurate
already
If you are using scalingType: deployment, the strategy is not related
@vietnd96 , We are using 'scalingType' for deployment. I upgraded the Selenium version from 4.16.1 to 4.17, but the issue persists. In a few days, we will be updating our KEDA and testing this issue. Meanwhile, please let me know if you can recommend any other fixes. I greatly appreciate your time and feedback.
@VietND96, We have now upgraded our infrastructure to the below versions. However, even after the upgrade, the issue remains the same. We still notice that some of our Edge nodes are going into the queue. Can you please assist on this issue.?
Kubernetes version: 1.27.7 KEDA: 2.12.1 Selenium grid: 4.18.1 Edge: 122 Chrome: 122
@VietND96 Can you please assist with this? Please let me know if you need any further information from my end.
Hi @chandupranayp, I will get back to you on this when having any clue. Besides this issue, also having some other unstable related to autoscaling are under investigation.
Hello @VietND96, thank you so much for the acknowledgment.
Hello @VietND96, any update on my issue, pls?
@chandupranayp, the exact root cause has yet to be identified. However, 2 fixes are available from the Grid server. https://github.com/SeleniumHQ/selenium/pull/14272 (delivered in 4.23) https://github.com/SeleniumHQ/selenium/pull/14282 (will be delivered in 4.23) We will continue to keep track of this issue.
FYI, image tag 4.23.0-20240727
and chart version 0.33.0
contain the fixes mentioned above. Kindly verify and provide feedback if it is the right fix for this issue.
This issue has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.
What happened?
After upgrading to Selenium version 4.16.1 and Edge 120, we have encountered an issue where some of the Edge nodes are being placed in a queue. Previously, we were using version 4.13.0 and Edge 117, and did not experience this problem. It seems that this issue is specific to Edge, as Chrome is functioning properly.
For example, when we trigger 5 Edge and 5 Chrome scripts, only 4 Edge nodes and all 5 Chrome nodes will open. One Edge node will be placed in the queue, despite setting the maxReplicaCount to 50.
Command used to start Selenium Grid with Docker (or Kubernetes)
Relevant log output
Operating System
Kubernetes version: 1.26.6
Docker Selenium version (image tag)
Selenium version 4.16.1 and Edge 120
Selenium Grid chart version (chart version)
No response