-
Notifications
You must be signed in to change notification settings - Fork 26
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 reources management for nodes #508
Conversation
b9e66d7
to
a99178c
Compare
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.
This still requires "documentation" in the chart's default values.yaml
What we typically do is provide a commented-out example in the values file, with some additional explanatory commentary.
25aa004
to
ee1e21f
Compare
4da46d5
to
8f984eb
Compare
"image" "octez" | ||
-}} | ||
|
||
{{- if hasKey $.node_vals "resources" }} |
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.
it looks like you can simply pass "resources" $.node.vals.resources
and if the key is absent it would pass null, and still work? So you can avoid setting $bakerParams at all and just modify the dict passed to include "tezos.generic_container"
.
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.
Why assign requests to the baker as well as the node? Then if you set requests of 3300Mi on a baking node, the pod actually requires 6600Mi. and more if you have more bakers. Can we set a request for the node only, so the value configured ends up being the total pod request?
d291af7
to
c56dfb6
Compare
@harryttd @orcutt989 all issues resolved, please review again |
118c71f
to
d5ae36b
Compare
@nicolasochem @harryttd There are still requested changes on this PR. Please dismiss them if they have been addressed. |
I'm reviewing right now |
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.
Almost ready. I think i understand the PR more fully now. I didn't get that the scaler wouldn't scale if it was targeting specific containers as opposed to the entire pod resources. So if it couldn't find any resources defined on any container it wouldn't scale.
Co-authored-by: Aryeh Harris <[email protected]>
This is a duplicated one of #502, PR generated from official repo branch to run all the CI jobs
This PR would set the resource manage for octez nodes and sidecar in the same pod. For bakers, there are no resources management for it however
Users can manage the node resources for octez node placement. Additionally, users can use k8s autoscalers to scale out node pods based on certain criteria
https://kubernetes.io/docs/tasks/run-application/horizontal-pod-autoscale
This scaling mechanism is for pod, so both octez node and sidecar resources should be specified for the pod resource usage to be calculated. For now, pods with bakers running inside is not supported.