-
Notifications
You must be signed in to change notification settings - Fork 1.1k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
fix: stop scale loop for pause Scaledbject (issue #4253) #4550
Conversation
@zroubalik @tomkerkhove Resubmitting a new PR. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Great, could you please update this PR to contain newest commits? there are some improvements in e2e tests logging, thanks!
/run-e2e |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks for the contribution, there are some minor nits and we should also introduce a new Condition in the ScaledObject.Status.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looking good, could you please add a marker to CRD to show PAUSED column and regenerate the CRD (make generate
)?
Probably before the AGE column:
https://github.com/kedacore/keda/blob/main/apis/keda/v1alpha1/scaledobject_types.go#L36
/run-e2e |
Sure thing. I added it and run |
/run-e2e |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
the e2e test for paused feature fails :(
ok thanks @zroubalik , I'll have a look at the pause_scaling_test logs, I guess this is what failed (I can't seem to find the details of the failure in the GitHub actions link) |
@zroubalik The e2e tests are passing in my cluster, but I suspect that the test needs more time to recover the scaling loop (loads might be inducing some more delay), so I'm increasing the wait time. Could you please trigger another run when you get a chance? Thanks |
/run-e2e |
/run-e2e pause |
That's weird. I'll need to have a deeper look later, as I'm at loss understanding the failures here. |
Seems like the testing deployment is not even initally scaled out, it stays on 0 replicas. |
Yes, that's exactly what is puzzling me. Need to sort it out! |
I made two changes to the test
|
/run-e2e pause |
@tobotg I ran your branch locally and got the same error as we can see here in the PR. |
@zroubalik Interesting that you're able to reproduce it locally, as I'm not able to reproduce it here. Any chance you can share Keda logs and app:pause-scaling-test and app:pause-scaling-test-monitored logs? That could be helpful for debugging. Thanks in advance. |
Signed-off-by: Tobo Atchou <tobo.atchou@gmail.com>
…or msg and new condition msg Signed-off-by: Tobo Atchou <tobo.atchou@gmail.com>
…sed condition Signed-off-by: Tobo Atchou <tobo.atchou@gmail.com>
… get hpa name and paused condition Signed-off-by: Tobo Atchou <tobo.atchou@gmail.com>
…ifest for paused condition Signed-off-by: Tobo Atchou <tobo.atchou@gmail.com>
…rease test wait timeout Signed-off-by: Tobo Atchou <tobo.atchou@gmail.com>
…ed up test with slim container image and wait for monitored deployment Signed-off-by: Tobo Atchou <tobo.atchou@gmail.com>
Signed-off-by: Tobo Atchou <tobo.atchou@gmail.com>
@zroubalik I had some time tonight to look into this issue, and submitted changes to the PR. Could you please trigger an e2e test when you get a chance? Thanks |
/run-e2e pause |
/run-e2e |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
/lgtm
Thanks @tobotg! Awesome job, it is a great improvement!
…acore#4550) Signed-off-by: Tobo Atchou <tobo.atchou@gmail.com>
In the controller in the reconcile loop, when a scaledObject is paused (annotated with ), this fix stops the scaling logic (to prevent attempt to connect the to scalers), and deletes the generated HPA (so it is not querying KEDA Metric Server for metrics)
Checklist
Fixes #4253