Repository navigation
Scale router/EPP resources for sim/cpu accelerator profiles - #4
Merged
Merged
Conversation
Fixes agentgateway#2. The router/EPP pod requests cpu: 8, memory: 8Gi by default (the llm-d-router chart's own value), sized for a real GPU node. Neither sim.yaml nor cpu.yaml set anything for router.epp, so agentgateway-standalone and agentgateway-gateway couldn't schedule on a laptop-sized Kind cluster - confirmed via kubectl describe showing "Insufficient cpu, Insufficient memory" on a fresh cluster. Added a router.epp.resources override to both profiles, matching the existing pattern already used for decode.resources. Verified the override reaches the rendered router-values.yaml correctly (render-scenario.py's router: block is a straight pass-through to the chart), and re-verified again after review. Signed-off-by: abhay1999 <abhaychaurasiya19@gmail.com>
danehans
approved these changes
Sep 17, 2026
Contributor
|
/cc @howardjohn |
howardjohn
approved these changes
Sep 17, 2026
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
3 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What this is
Fixes #2.
The router/EPP pod requests
cpu: 8, memory: 8Giby default (the llm-d-router chart's own value, sized for a real GPU node). Neithersim.yamlnorcpu.yamlset anything underrouter.epp, soagentgateway-standaloneandagentgateway-gatewaycouldn't schedule on a laptop-sized Kind cluster at all:Fix
Added a
router.epp.resourcesoverride to bothsim.yamlandcpu.yaml, matching the existing pattern already used fordecode.resourcesin the same files.Verification
render-scenario.py'srouter:block is a straight pass-through to the llm-d-router chart's values (see12_router-values.yaml.j2's own comment), so anything set here reaches the chart directly.agentgateway-standalonescenario against a real Kind cluster and confirmed the resultingrouter-values.yamlshows the newresourcesblock correctly.ghcr.io/llm-d/llm-d-benchmark:nightlycurrently has no arm64 image manifest, which fails the harness pod before the run ever reaches the router pod. That's an unrelated upstream issue, not something this PR touches.