l4proxy: add Prometheus metrics#431
Open
tannevaled wants to merge 2 commits into
Open
Conversation
The layer4 proxy exposed no metrics. Add per-upstream collectors: - caddy_layer4_proxy_connections_total (counter) - caddy_layer4_proxy_active_connections (gauge) - caddy_layer4_proxy_upstream_healthy (gauge, from active health checks) Collectors are registered on the instance metrics registry obtained from the Caddy context (ctx.GetMetricsRegistry), the same mechanism the HTTP server uses, so they reset cleanly across config reloads and surface on the existing admin /metrics endpoint. Metric updates are nil-safe so a handler that was never provisioned (e.g. in tests) does not panic. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Owner
|
Thank you for all your contributions lately. I imagine you're using an LLM to assist? We will look at them soon! |
Author
|
Hi @mholt — thanks for the kind note! Yes, co-authored with Claude Opus 4.8. I design, test, and review every change myself before submitting. These PRs implement features I'm missing downstream, so the work is purposeful and scoped. Happy to slow down if the review load is a concern — no rush on your side! |
The metric helpers are already at 100% coverage; this adds a Metrics section to docs/handlers/proxy.md listing the exposed collectors. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
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
2 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
Adds Prometheus metrics to the layer4 proxy, which previously exposed none:
caddy_layer4_proxy_connections_total{upstream}— counter of proxied connectionscaddy_layer4_proxy_active_connections{upstream}— gauge of in-flight proxied connectionscaddy_layer4_proxy_upstream_healthy{upstream}— gauge (1/0) reflecting active health-check stateWhy
Observability is the one big thing the L4 proxy lacked versus more mature TCP load balancers — no way to see connection counts or upstream health from metrics. These three cover the basics (load + health) per upstream.
Notes (re: de-duplication)
This deliberately does not copy the HTTP server's metrics code — it uses the same mechanism (
ctx.GetMetricsRegistry()+promauto.With(registry)), so collectors live on the per-instance registry, reset cleanly across reloads, and surface on the existing admin/metricsendpoint. Updates are nil-safe (a never-provisioned handler doesn't panic). Labels are kept to a singleupstreamto bound cardinality; happy to adjust names/labels to your conventions.Tests
metrics_test.go: counter/gauge increments and the open/close lifecycle, the health gauge, nil-safety, and that an active health check sets the health gauge to 0 (dead peer) and 1 (live listener).go test ./modules/l4proxy/passes;gofmt/go vet/golangci-lintclean. go.mod change is just promotingprometheus/client_golangfrom indirect to direct.