fix(metrics): update metrics recommendations and add OpenTelemetry review documentation

This commit is contained in:
Marc Schäfer
2025-10-10 18:18:38 +02:00
parent 1a9f6c4685
commit 46384e6242
3 changed files with 161 additions and 17 deletions

View File

@@ -10,6 +10,10 @@ This document captures the current state of Newt metrics, prioritized fixes, and
- Tunnel/Traffic: newt_tunnel_sessions, newt_tunnel_bytes_total, newt_tunnel_latency_seconds, newt_tunnel_reconnects_total
- Connection lifecycle: newt_connection_attempts_total, newt_connection_errors_total
- Operations: newt_config_reloads_total, newt_restart_count_total, newt_build_info
- Operations: newt_config_reloads_total, newt_restart_count_total, newt_config_apply_seconds, newt_cert_rotation_total
- Build metadata: newt_build_info
- Control plane: newt_websocket_connect_latency_seconds, newt_websocket_messages_total
- Proxy: newt_proxy_active_connections, newt_proxy_buffer_bytes, newt_proxy_async_backlog_bytes, newt_proxy_drops_total
- Go runtime: GC, heap, goroutines via runtime instrumentation
2) Main issues addressed now
@@ -27,6 +31,10 @@ This document captures the current state of Newt metrics, prioritized fixes, and
- Some call sites still need initiator label on reconnect outcomes (client vs server). This is planned.
- WebSocket and Proxy metrics (connect latency, messages, active connections, buffer/drops, async backlog) are planned additions.
- Config apply duration and cert rotation counters are planned.
- Registration and config reload failures are not yet emitted; add failure code paths so result labels expose churn.
- Restart counter increments only when build metadata is provided; consider decoupling to count all boots.
- Metric helpers often use `context.Background()`. Where lightweight contexts exist (e.g., HTTP handlers), propagate them to ease future correlation.
- Tracing coverage is limited to admin HTTP and WebSocket connect spans; extend to blueprint fetches, proxy accept loops, and WireGuard updates when OTLP is enabled.
4) Roadmap (phased)
@@ -40,6 +48,10 @@ This document captures the current state of Newt metrics, prioritized fixes, and
- Proxy: newt_proxy_active_connections, newt_proxy_buffer_bytes, newt_proxy_drops_total, newt_proxy_async_backlog_bytes
- Reconnect: add initiator label (client/server)
- Config & PKI: newt_config_apply_seconds{phase,result}; newt_cert_rotation_total{result}
- WebSocket disconnect and keepalive failure counters
- Proxy connection lifecycle metrics (accept totals, duration histogram)
- Pangolin blueprint/config fetch latency and status metrics
- Certificate rotation duration histogram to complement success/failure counter
5) Operational guidance
@@ -64,9 +76,3 @@ This document captures the current state of Newt metrics, prioritized fixes, and
- Direct scrape variant requires no attribute promotion since site_id is already a metric label.
- Transform/promote variant remains optional for environments that rely on resource-to-label promotion.
8) Testing
- curl :2112/metrics | grep ^newt_
- Verify presence of site_id across series; region appears when set.
- Ensure disallowed attributes are filtered; allowed (site_id) retained.

View File

@@ -34,18 +34,30 @@ Runtime behavior
- When OTLP is enabled, metrics and traces are exported to OTLP gRPC endpoint
- Go runtime metrics (goroutines, GC, memory) are exported automatically
Metric catalog (initial)
Metric catalog (current)
- newt_build_info (gauge) labels: version, commit, site_id[, region]; value is always 1
- newt_site_registrations_total (counter) labels: result, site_id[, region]
- newt_site_online (observable gauge) labels: site_id (0/1)
- newt_site_last_heartbeat_seconds (observable gauge) labels: site_id
- newt_tunnel_sessions (observable gauge) labels: site_id, tunnel_id [transport optional when known]
- newt_tunnel_bytes_total (counter) labels: site_id, tunnel_id, protocol (tcp|udp), direction (ingress|egress)
- newt_tunnel_latency_seconds (histogram) labels: site_id, tunnel_id, transport (e.g., wireguard)
- newt_tunnel_reconnects_total (counter) labels: site_id, tunnel_id, initiator (client|server), reason
- newt_connection_attempts_total (counter) labels: site_id, transport, result
- newt_connection_errors_total (counter) labels: site_id, transport, error_type (dial_timeout|tls_handshake|auth_failed|io_error)
| Metric | Instrument | Key attributes | Purpose | Example |
| --- | --- | --- | --- | --- |
| `newt_build_info` | Observable gauge (Int64) | `version`, `commit`, `site_id`, `region` (optional) | Emits build metadata with value `1` for scrape-time verification. | `newt_build_info{version="1.5.0",site_id="acme-edge-1"} 1` |
| `newt_site_registrations_total` | Counter (Int64) | `result` (`success`/`failure`), `site_id`, `region` (optional) | Counts Pangolin registration attempts. | `newt_site_registrations_total{result="success",site_id="acme-edge-1"} 1` |
| `newt_site_online` | Observable gauge (Int64) | `site_id` | Reports whether the site is currently connected (`1`) or offline (`0`). | `newt_site_online{site_id="acme-edge-1"} 1` |
| `newt_site_last_heartbeat_seconds` | Observable gauge (Float64) | `site_id` | Time since the most recent Pangolin heartbeat. | `newt_site_last_heartbeat_seconds{site_id="acme-edge-1"} 2.4` |
| `newt_tunnel_sessions` | Observable gauge (Int64) | `site_id`, `tunnel_id` (when enabled) | Counts active tunnel sessions per peer; collapses to per-site when tunnel IDs are disabled. | `newt_tunnel_sessions{site_id="acme-edge-1",tunnel_id="wgpub..."} 3` |
| `newt_tunnel_bytes_total` | Counter (Int64) | `direction` (`ingress`/`egress`), `protocol` (`tcp`/`udp`), `tunnel_id` (optional), `site_id`, `region` (optional) | Measures proxied traffic volume across tunnels. | `newt_tunnel_bytes_total{direction="ingress",protocol="tcp",site_id="acme-edge-1"} 4096` |
| `newt_tunnel_latency_seconds` | Histogram (Float64) | `transport` (e.g., `wireguard`), `tunnel_id` (optional), `site_id`, `region` (optional) | Captures RTT or configuration-driven latency samples. | `newt_tunnel_latency_seconds_bucket{transport="wireguard",le="0.5"} 42` |
| `newt_tunnel_reconnects_total` | Counter (Int64) | `initiator` (`client`/`server`), `reason` (enumerated), `tunnel_id` (optional), `site_id`, `region` (optional) | Tracks reconnect causes for troubleshooting flaps. | `newt_tunnel_reconnects_total{initiator="client",reason="timeout",site_id="acme-edge-1"} 5` |
| `newt_connection_attempts_total` | Counter (Int64) | `transport` (`auth`/`websocket`), `result`, `site_id`, `region` (optional) | Measures control-plane dial attempts and their outcomes. | `newt_connection_attempts_total{transport="websocket",result="success",site_id="acme-edge-1"} 8` |
| `newt_connection_errors_total` | Counter (Int64) | `transport`, `error_type`, `site_id`, `region` (optional) | Buckets connection failures by normalized error class. | `newt_connection_errors_total{transport="websocket",error_type="tls_handshake",site_id="acme-edge-1"} 1` |
| `newt_config_reloads_total` | Counter (Int64) | `result`, `site_id`, `region` (optional) | Counts remote blueprint/config reloads. | `newt_config_reloads_total{result="success",site_id="acme-edge-1"} 3` |
| `newt_restart_count_total` | Counter (Int64) | `site_id`, `region` (optional) | Increments once per process boot to detect restarts. | `newt_restart_count_total{site_id="acme-edge-1"} 1` |
| `newt_config_apply_seconds` | Histogram (Float64) | `phase` (`interface`/`peer`), `result`, `site_id`, `region` (optional) | Measures time spent applying WireGuard configuration phases. | `newt_config_apply_seconds_sum{phase="peer",result="success",site_id="acme-edge-1"} 0.48` |
| `newt_cert_rotation_total` | Counter (Int64) | `result`, `site_id`, `region` (optional) | Tracks client certificate rotation attempts. | `newt_cert_rotation_total{result="success",site_id="acme-edge-1"} 2` |
| `newt_websocket_connect_latency_seconds` | Histogram (Float64) | `transport="websocket"`, `result`, `error_type` (on failure), `site_id`, `region` (optional) | Measures WebSocket dial latency and exposes failure buckets. | `newt_websocket_connect_latency_seconds_bucket{result="success",le="0.5",site_id="acme-edge-1"} 9` |
| `newt_websocket_messages_total` | Counter (Int64) | `direction` (`in`/`out`), `msg_type` (`text`/`ping`/`pong`), `site_id`, `region` (optional) | Accounts for control WebSocket traffic volume by type. | `newt_websocket_messages_total{direction="out",msg_type="ping",site_id="acme-edge-1"} 12` |
| `newt_proxy_active_connections` | Observable gauge (Int64) | `protocol` (`tcp`/`udp`), `direction` (`ingress`/`egress`), `tunnel_id` (optional), `site_id`, `region` (optional) | Current proxy connections per tunnel and protocol. | `newt_proxy_active_connections{protocol="tcp",direction="egress",site_id="acme-edge-1"} 4` |
| `newt_proxy_buffer_bytes` | Observable gauge (Int64) | `protocol`, `direction`, `tunnel_id` (optional), `site_id`, `region` (optional) | Volume of buffered data awaiting flush in proxy queues. | `newt_proxy_buffer_bytes{protocol="udp",direction="egress",site_id="acme-edge-1"} 2048` |
| `newt_proxy_async_backlog_bytes` | Observable gauge (Int64) | `protocol`, `direction`, `tunnel_id` (optional), `site_id`, `region` (optional) | Tracks async write backlog when deferred flushing is enabled. | `newt_proxy_async_backlog_bytes{protocol="tcp",direction="egress",site_id="acme-edge-1"} 512` |
| `newt_proxy_drops_total` | Counter (Int64) | `protocol`, `tunnel_id` (optional), `site_id`, `region` (optional) | Counts proxy drop events caused by downstream write errors. | `newt_proxy_drops_total{protocol="udp",site_id="acme-edge-1"} 1` |
Conventions

126
docs/otel-review.md Normal file
View File

@@ -0,0 +1,126 @@
# Newt OpenTelemetry Review
## Overview
This document summarises the current OpenTelemetry (OTel) instrumentation in Newt, assesses
compliance with OTel guidelines, and lists concrete improvements to pursue before release.
It is based on the implementation in `internal/telemetry` and the call-sites that emit
metrics and traces across the code base.
## Current metric instrumentation
All instruments are registered in `internal/telemetry/metrics.go`. They are grouped
into site, tunnel, connection, configuration, build, WebSocket, and proxy domains.
A global attribute filter (see `buildMeterProvider`) constrains exposed label keys to
`site_id`, `region`, and a curated list of low-cardinality dimensions so that Prometheus
exports stay bounded.
- **Site lifecycle**: `newt_site_registrations_total`, `newt_site_online`, and
`newt_site_last_heartbeat_seconds` capture registration attempts and liveness. They
are fed either manually (`IncSiteRegistration`) or via the `TelemetryView` state
callback that publishes observable gauges for the active site.
- **Tunnel health and usage**: Counters and histograms track bytes, latency, reconnects,
and active sessions per tunnel (`newt_tunnel_*` family). Attribute helpers respect
the `NEWT_METRICS_INCLUDE_TUNNEL_ID` toggle to keep cardinality manageable on larger
fleets.
- **Connection attempts**: `newt_connection_attempts_total` and
`newt_connection_errors_total` are emitted throughout the WebSocket client to classify
authentication, dial, and transport failures.
- **Operations/configuration**: `newt_config_reloads_total`,
`newt_restart_count_total`, `newt_config_apply_seconds`, and
`newt_cert_rotation_total` provide visibility into blueprint reloads, process boots,
configuration timings, and certificate rotation outcomes.
- **Build metadata**: `newt_build_info` records the binary version/commit together
with a monotonic restart counter when build information is supplied at startup.
- **WebSocket control-plane**: `newt_websocket_connect_latency_seconds` and
`newt_websocket_messages_total` report connect latency and ping/pong/text activity.
- **Proxy data-plane**: Observable gauges (`newt_proxy_active_connections`,
`newt_proxy_buffer_bytes`, `newt_proxy_async_backlog_bytes`) and the
`newt_proxy_drops_total` counter are fed from the proxy manager to monitor backlog
and drop behaviour alongside per-protocol byte counters.
Refer to `docs/observability.md` for a tabular catalogue with instrument types,
attributes, and sample exposition lines.
## Tracing coverage
Tracing is optional and enabled only when OTLP export is configured. When active:
- The admin HTTP mux is wrapped with `otelhttp.NewHandler`, producing spans for
`/metrics` and `/healthz` requests.
- The WebSocket dial path creates a `ws.connect` span around the gRPC-based handshake.
No other subsystems currently create spans, so data-plane operations, blueprint fetches,
Docker discovery, and WireGuard reconfiguration happen without trace context.
## Guideline & best-practice alignment
The implementation adheres to most OTel Go recommendations:
- **Naming & units** Every instrument follows the `newt_*` prefix with `_total`
suffixes for counters and `_seconds`/`_bytes` unit conventions. Histograms are
registered with explicit second-based buckets.
- **Resource attributes** Service name/version and optional `site_id`/`region`
populate the `resource.Resource` and are also injected as metric attributes for
compatibility with Prometheus queries.
- **Attribute hygiene** A single attribute filter (`sdkmetric.WithView`) enforces
the allow-list of label keys to prevent accidental high-cardinality emission.
- **Runtime metrics** Go runtime instrumentation is enabled automatically through
`runtime.Start`.
- **Configuration via environment** `telemetry.FromEnv` honours `OTEL_*` variables
alongside `NEWT_*` overrides so operators can configure exporters without code
changes.
- **Shutdown handling** `Setup.Shutdown` iterates exporters in reverse order to
flush buffers before process exit.
## Adjustments & improvements
The review identified a few actionable adjustments:
1. **Record registration failures** `newt_site_registrations_total` is currently
incremented only on success. Emit `result="failure"` samples whenever Pangolin
rejects a registration or credential exchange so operators can alert on churn.
2. **Surface config reload failures** `telemetry.IncConfigReload` is invoked with
`result="success"` only. Callers should record a failure result when blueprint
parsing or application aborts before success counters are incremented.
3. **Harmonise restart count behaviour** `newt_restart_count_total` increments only
when build metadata is provided. Consider moving the increment out of
`RegisterBuildInfo` so the counter advances even for ad-hoc builds without version
strings.
4. **Propagate contexts where available** Many emitters call metric helpers with
`context.Background()`. Passing real contexts (when inexpensive) would allow future
exporters to correlate spans and metrics.
5. **Extend tracing coverage** Instrument critical flows such as blueprint fetches,
WireGuard reconfiguration, proxy accept loops, and Docker discovery to provide end
to end visibility when OTLP tracing is enabled.
## Metrics to add before release
Prioritised additions that would close visibility gaps:
1. **WebSocket disconnect outcomes** A counter (e.g., `newt_websocket_disconnects_total`)
partitioned by `reason` would complement the existing connect latency histogram and
explain reconnect storms.
2. **Keepalive/heartbeat failures** Counting ping timeouts or heartbeat misses would
make `newt_site_last_heartbeat_seconds` actionable by providing discrete events.
3. **Proxy connection lifecycle** Add counters/histograms for proxy accept events and
connection durations to correlate drops with load and backlog metrics.
4. **Blueprint/config pull latency** Measuring Pangolin blueprint fetch durations and
HTTP status distribution would expose slow control-plane operations.
5. **Certificate rotation attempts** Complement `newt_cert_rotation_total` with a
duration histogram to observe slow PKI updates and detect stuck rotations.
These metrics rely on data that is already available in the code paths mentioned
above and would round out operational dashboards.
## Tracing wishlist
To benefit from tracing when OTLP is active, add spans around:
- Pangolin REST calls (wrap the HTTP client with `otelhttp.NewTransport`).
- Docker discovery cycles and target registration callbacks.
- WireGuard reconfiguration (interface bring-up, peer updates).
- Proxy dial/accept loops for both TCP and UDP targets.
Capturing these stages will let operators correlate latency spikes with reconnects
and proxy drops using distributed traces in addition to the metric signals.