fix(gateway): carry a remote node's served name into its wake stats - #235
Merged
Merged
Conversation
The stats Lambda already read the deploy config's servedModelName but never included it in its reply, so a stopped remote node's wake match only had its bare model id to compare against. A caller that had been serving requests under the node's ALIAS while it was running lost that match the moment it stopped, and the gateway refused to wake it.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fixes a mismatch where a remote node served a request under its ALIAS while running, but the gateway refused to wake it under that same name once it stopped.
Summary
servedModelNamebut never put it on its reply, so a stopped remote node's wake match only had the bare model id to compare against.remote/lambda/shared/stats.ts/remote/lambda/stats/index.ts: the stats Lambda now includesservedNamefrom the deploy config, in both the running and stopped/undeployed reply branches.internal/remote/remote.go:StatsResponsegains aServedNamefield to receive it.internal/fleet/remote_node.go:statsFromRemotecarries it intometrics.Stats.internal/metrics/metrics.go/internal/daemon/daemon.go:Statsgains aServedNamefield, and the local daemon's ownMetrics()now fills it too — it already tracked the served name internally but was dropping it here.internal/gateway/gateway.go:remoteConfigFornow returnsServedModelNamefrom the stats reply, so wake-matching and/v1/modelsuse the same served-name-first naming a running node already reports.docs/openapi.yaml: documents the new field.Implementation details
This requires redeploying the stats Lambda (
remote/lambda/stats) to take effect for existing remote environments — the Go-side gateway changes alone are not enough, since the field has to actually be present on the wire.