Collect, normalize, and aggregate fleet metrics - #380
Closed
dennis-upbound wants to merge 1 commit into
Closed
Conversation
The metrics design normalizes each engine's Prometheus metrics onto a modelplane_* surface, selected per engine by a first-class resource rather than a hand-edited ConfigMap. Add that resource: a cluster-scoped MetricMapping XRD carrying a pod selector, a rename map, and label rewrites, plus a mark-ready composition function, following the InferenceClass config-kind pattern. The collector that reads these and the built-in per-engine mappings follow in later commits. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Signed-off-by: Dennis Ramdass <dennis@upbound.io>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Implements the metrics design (#363, addressing #269): collect on every cluster, normalize to
modelplane_*, and aggregate to one control-plane view through an OpenTelemetry collector.Draft, built up in commits. This first commit adds the
MetricMappingkind — the first-class, cluster-scoped resource that carries an engine's metric rename and label rules, which the collector reads to normalize ontomodelplane_*. It follows theInferenceClassconfig-kind pattern: an XRD plus a mark-ready composition function (no composed children).Still to come on this branch:
MetricMappings for vLLM, SGLang, and Triton/TensorRT-LLM.compose-serving-stack: the per-cluster OTel collector — scrape themodelplane.ai/servingselector plus the EPP and substrate, transform-rename via theMetricMappings, push outbound./metricsporthttpinnative.py,llmd.py,routing.py.Adding a kind requires
nix run .#buildto regenerate the models; CI runs it.