Repository navigation
Conversation
google#663 raised the router pod's CPU request to 500m. hostPort 4501 pins one router per node, the hub's routers hold three of the ten nodes' ports, and the one node left for agentmesh-router-bananas-4 has 166m of CPU unrequested, so the pod stayed Pending and the deploy of 4c8124b waited on the rollout. The request is 100m again; the 1 GiB memory request fits and stays.
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Code Review
This pull request updates the Kubernetes template agentmesh-router-template.yaml to reduce the CPU request for the router from 500m to 100m. This change, along with an updated explanatory comment, prevents the router from remaining in a Pending state on nodes where CPU resources are limited due to port pinning. There are no review comments, and I have no additional feedback to provide.
Collaborator
Author
|
Superseded: the default pool was grown from 3 to 4 nodes per zone (9 → 12 nodes). agentmesh-router-bananas-4 scheduled on a new node with the 500m request and the rollout of 4c8124b resumed; with four spare nodes the request a router needs for its handshakes can stand. |
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
#663 raised the router pod's CPU request from 100m to 500m.
hostPort 4501pins one router per node; the hub's routers hold three of the ten nodes' ports, and the only node left foragentmesh-router-bananas-4has 166m of CPU unrequested (twoopenroutercanaries at 600m each), so the pod stayed Pending (0/10 nodes are available: 1 Insufficient cpu, 8 node(s) didn't have free ports) and the deploy of4c8124b4sat on the router rollout. The request is 100m again; the 1 GiB memory request fits (that node has 28% of memory requested) and stays.