-
Notifications
You must be signed in to change notification settings - Fork 4k
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
Add support for scaling up with ZeroToMaxNodesScaling option #5826
Conversation
So the purpose of this feature is we just want to scale up to the max when the flag is set to true? How should scale down be defined for this feature? What is the condition for atomic scale up? Just for a normal scale operation but we scale the entire nodepool up to max count? |
Correct, when the flag is set even if only a single node is sufficient for the pod we scale up to the max.
Please take a look at #5695
I don't quite understand the question, it is like a normal scale operation only difference is if the node group has the atomic option we scale up the whole group. |
cluster-autoscaler/core/scaleup/orchestrator/orchestrator_test.go
Outdated
Show resolved
Hide resolved
e98c397
to
639253e
Compare
Can we rename this mechanism? My expectation for 'atomic scaling' would be along the lines of handling a set of pods (perhaps a job?) in a single scale-up operation on a single NodeGroup or something along those lines. This is very far from what I'd expect and I can easily imagine someone enabling this feature based on their intuition around the name of the flag and getting very surprised with a large bill after their nodegroup scaled to max. Maybe something like MaxOrNothingScaling? |
I like it, the solution is much more elegant now. Great work! |
Renamed the mechanism to |
/lgtm |
/approve |
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: hbostan, MaciekPytel The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
/lgtm |
/lgtm |
* Merged multiple tests into one single table driven test. * Fixed some typos.
…strator * Started handling scale up options for ZeroToMaxNodeScaling with the existing estimator * Skip setting similar node groups for the node groups that use ZeroToMaxNodeScaling * Renamed the autoscaling option from "AtomicScaleUp" to "AtomicScaling" * Merged multiple tests into one single table driven test. * Fixed some typos.
* Renamed the "AtomicScaling" autoscaling option to "ZeroOrMaxNodeScaling" to be more clear about the behavior.
/lgtm |
/unhold |
What type of PR is this?
/kind feature
What this PR does / why we need it:
This PR adds a node group AtomicScaleUp option, that allows for all-or-nothing scale up of the node group.
Which issue(s) this PR fixes:
N/A
Special notes for your reviewer:
Does this PR introduce a user-facing change?