mirror of
https://github.com/rustfs/rustfs.git
synced 2026-08-06 05:17:42 +00:00
houseme
25c6bdf490
* chore(perf): harden amd64 profiling benchmark flow * fix(profiling): isolate bench buckets and map protobuf conflict * perf: avoid blocking owned local writes * style: format profile admin handler * docs: clarify observability trace validation * perf: reduce mkdir overhead on local writes * perf: add rename_data meta microbenchmark * perf(filemeta): fast-path data_dir decode in version meta * perf(filemeta): collapse data-dir lookup into one scan * perf(filemeta): reduce scan allocs and refresh meta bench * perf(ecstore): skip mkdir path on read-only open * perf(filemeta): single-pass unshared data-dir scan * perf(filemeta): add two-key inline remove fast path * perf(filemeta): compare remove-two keys by bytes first * bench(ecstore): add remove_two-only micro benchmark * bench(ecstore): stabilize rename_data meta benchmark timing * bench(ecstore): align rename_data path with remove_two * perf(filemeta): avoid uuid string alloc in remove_two * perf(filemeta): add fast-path for empty inline data * perf(filemeta): streamline add_version match branch * perf(filemeta): fast-return remove_key on miss * perf(filemeta): speed up add_version insertion lookup * style(ecstore): normalize formatting in perf-tuning files * refactor(filemeta): unify inline data removal paths
146 lines
5.6 KiB
Markdown
146 lines
5.6 KiB
Markdown
# RustFS Observability Stack
|
|
|
|
This directory contains the comprehensive observability stack for RustFS, designed to provide deep insights into application performance, logs, and traces.
|
|
|
|
## Components
|
|
|
|
The stack is composed of the following best-in-class open-source components:
|
|
|
|
- **Prometheus** (v2.53.1): The industry standard for metric collection and alerting.
|
|
- **Grafana** (v11.1.0): The leading platform for observability visualization.
|
|
- **Loki** (v3.1.0): A horizontally-scalable, highly-available, multi-tenant log aggregation system.
|
|
- **Tempo** (v2.5.0): A high-volume, minimal dependency distributed tracing backend.
|
|
- **Jaeger** (v1.59.0): Distributed tracing system (configured as a secondary UI/storage).
|
|
- **OpenTelemetry Collector** (v0.104.0): A vendor-agnostic implementation for receiving, processing, and exporting telemetry data.
|
|
|
|
By default, this stack uses Tempo in single-binary mode and does not require Kafka/Redpanda.
|
|
If you want the Kafka-backed HA Tempo path, use `docker-compose-example-for-rustfs.yml` together with `docker-compose-tempo-ha-override.yml`.
|
|
|
|
## Architecture
|
|
|
|
1. **Telemetry Collection**: Applications send OTLP (OpenTelemetry Protocol) data (Metrics, Logs, Traces) to the **OpenTelemetry Collector**.
|
|
2. **Processing & Exporting**: The Collector processes the data (batching, memory limiting) and exports it to the respective backends:
|
|
- **Traces** -> **Tempo** (Primary) & **Jaeger** (Secondary/Optional)
|
|
- **Metrics** -> **Prometheus** (via scraping the Collector's exporter)
|
|
- **Logs** -> **Loki**
|
|
3. **Visualization**: **Grafana** connects to all backends (Prometheus, Tempo, Loki, Jaeger) to provide a unified dashboard experience.
|
|
|
|
## Features
|
|
|
|
- **Full Persistence**: All data (Metrics, Logs, Traces) is persisted to Docker volumes, ensuring no data loss on restart.
|
|
- **Correlation**: Seamless navigation between Metrics, Logs, and Traces in Grafana.
|
|
- Jump from a Metric spike to relevant Traces.
|
|
- Jump from a Trace to relevant Logs.
|
|
- **High Performance**: Optimized configurations for batching, compression, and memory management.
|
|
- **Standardized Protocols**: Built entirely on OpenTelemetry standards.
|
|
|
|
## Quick Start
|
|
|
|
### Prerequisites
|
|
|
|
- Docker
|
|
- Docker Compose
|
|
|
|
### Deploy
|
|
|
|
Run the following command to start the entire stack:
|
|
|
|
```bash
|
|
docker compose up -d
|
|
```
|
|
|
|
### High Availability Tempo
|
|
|
|
The default `docker-compose.yml` is the single-node stack.
|
|
If you need the Kafka-backed HA Tempo configuration, start it with:
|
|
|
|
```bash
|
|
docker compose -f docker-compose-example-for-rustfs.yml -f docker-compose-tempo-ha-override.yml up -d
|
|
```
|
|
|
|
### Access Dashboards
|
|
|
|
| Service | URL | Credentials | Description |
|
|
| :------------- | :----------------------------------------------- | :---------------- | :----------------------------- |
|
|
| **Grafana** | [http://localhost:3000](http://localhost:3000) | `admin` / `admin` | Main visualization hub. |
|
|
| **Prometheus** | [http://localhost:9090](http://localhost:9090) | - | Metric queries and status. |
|
|
| **Jaeger UI** | [http://localhost:16686](http://localhost:16686) | - | Secondary trace visualization. |
|
|
| **Tempo** | [http://localhost:3200](http://localhost:3200) | - | Tempo status/metrics. |
|
|
|
|
## Configuration
|
|
|
|
### Data Persistence
|
|
|
|
Data is stored in the following Docker volumes:
|
|
|
|
- `prometheus-data`: Prometheus metrics
|
|
- `tempo-data`: Tempo traces (WAL and Blocks)
|
|
- `loki-data`: Loki logs (Chunks and Rules)
|
|
- `jaeger-data`: Jaeger traces (Badger DB)
|
|
|
|
To clear all data:
|
|
|
|
```bash
|
|
docker compose down -v
|
|
```
|
|
|
|
### Customization
|
|
|
|
- **Prometheus**: Edit `prometheus.yml` to add scrape targets or alerting rules.
|
|
- **Grafana**: Dashboards and datasources are provisioned from the `grafana/` directory.
|
|
- **Collector**: Edit `otel-collector-config.yaml` to modify pipelines, processors, or exporters.
|
|
|
|
### Verifying RustFS Traces
|
|
|
|
When RustFS points `RUSTFS_OBS_ENDPOINT` at this stack, treat the value as the
|
|
OTLP/HTTP base URL, for example:
|
|
|
|
```bash
|
|
export RUSTFS_OBS_ENDPOINT=http://host.docker.internal:4318
|
|
```
|
|
|
|
RustFS automatically expands that base URL to:
|
|
|
|
- `/v1/traces`
|
|
- `/v1/metrics`
|
|
- `/v1/logs`
|
|
|
|
Important behavior notes:
|
|
|
|
- Logs and metrics usually appear during startup, so seeing those two signals
|
|
first is expected.
|
|
- Visible trace data usually requires real HTTP/S3/gRPC request traffic after
|
|
startup, because request-path spans are created on demand.
|
|
- `RUSTFS_OBS_LOGGER_LEVEL=info` keeps the top-level request span but filters
|
|
many nested `debug` spans. If Tempo or Jaeger looks sparse, retry with
|
|
`RUSTFS_OBS_LOGGER_LEVEL=debug` before suspecting collector or Tempo issues.
|
|
|
|
Minimal validation flow:
|
|
|
|
```bash
|
|
# 1. Start this observability stack.
|
|
docker compose up -d
|
|
|
|
# 2. Start RustFS with OTLP/HTTP export and richer span visibility.
|
|
export RUSTFS_OBS_ENDPOINT=http://host.docker.internal:4318
|
|
export RUSTFS_OBS_LOGGER_LEVEL=debug
|
|
|
|
# 3. Generate real request traffic.
|
|
curl -I http://127.0.0.1:9000/health
|
|
curl -I http://127.0.0.1:9000/health/ready
|
|
|
|
# 4. Inspect Grafana or Jaeger.
|
|
# Grafana: http://localhost:3000
|
|
# Jaeger: http://localhost:16686
|
|
```
|
|
|
|
If logs and metrics are present but traces are sparse, the most common cause is
|
|
"no real request traffic yet" or "`info` level filtered nested spans", not an
|
|
OTLP routing failure.
|
|
|
|
## Troubleshooting
|
|
|
|
- **Service Health**: Check the health of services using `docker compose ps`.
|
|
- **Logs**: View logs for a specific service using `docker compose logs -f <service_name>`.
|
|
- **Otel Collector**: Check `http://localhost:13133` for health status and `http://localhost:1888/debug/pprof/` for profiling.
|