Defer cEOS-lab pod check, update operator version #534
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
The Arista node is waiting for its pod to come up in Create(). This is problematic in high-scale scenarious because we create the pods sychronously. Move the check to Status() so checkNodeStatus() handles things instead.
See release notes: https://github.com/aristanetworks/arista-ceoslab-operator/releases/tag/v2.1.2 Increased the number of workers in the operator to the number of cores (from the default value of 1). This may improve performance in high-scale scenarios. Most of their time is spent generating RSA certificates.
Pull Request Test Coverage Report for Build 8963970828Details
💛 - Coveralls |
|
As the concerned customer, thanks for this. FWIW, I do have a hack that lets KNE avoid the local wait by launching all pods of a topology in parallel, with obvious concerns about simply shifting the problem to other API's -- in this case, that's what made apparent that the cEOS operator was itself somewhat serializing the work. |
|
/gcbrun |
alexmasi
approved these changes
May 6, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
A customer is expressing performance concerns at high scale (~100 instances across ~10 nodes). One of their findings is that cEOS-lab instances appear to start consecutively instead of in parallel.
Because the pod check is baked into
(n *Node) Configinstead of(n *Node) Status, we don't create the next cEOS-lab custom resource object until the previous pod has started. Now they're created all at once.The new operator version increases the number of reconcilation workers from 1 to
runtime.NumCPUto cope with this change. It turns out the operator spends most of its time generated self-signed RSA certs, depending on what the runtime does with the worker goroutines there may be performance gains there.Thanks!