mirror of https://github.com/fleetdm/fleet synced 2026-04-21 13:37:30 +00:00

History

Konstantin Sykulev 6ed3ba6801 Added OTEL DB stats metrics, renamed trace attributes to expected OTEL names (#42097 ) 1. Added DB metrics via otelsql.RegisterDBStatsMetrics() `db.sql.connection.open` `db.sql.connection.max_open` `db.sql.connection.wait` `db.sql.connection.wait_duration` `db.sql.connection.closed_max_idle` `db.sql.connection.closed_max_idle_time` `db.sql.latency.` 2. renamed these metrics to signoz convention/expected names `db.sql.connection.open` -> `db.client.connection.usage` `db.sql.connection.max_open` -> `db.client.connection.max` `db.sql.connection.wait` -> `db.client.connection.wait_count` `db.sql.connection.wait_duration` -> `db.client.connection.wait_time` `db.sql.connection.closed_max_idle` -> `db.client.connection.idle.max` `db.sql.connection.closed_max_idle_time` -> `db.client.connection.idle.min` 3. created custom dashboard to display these metrics, (import via json) <img width="1580" height="906" alt="Screenshot 2026-03-19 at 2 44 43 PM" src="https://github.com/user-attachments/assets/f1b64ed6-e534-4490-8955-bc1205dd21d4" /> 4. Fixed metrics for service db dashboards Signoz expects `db.system` : Identifies the database type (e.g., postgresql, mysql, mongodb). `db.statement` : The actual query being executed (e.g., SELECT FROM users). `db.operation` : The type of operation (e.g., SELECT, INSERT). `service.name` : The name of the service making the call. We needed to set the `db.system` attribute explicitly. `db.operation` is missing because otelsql doesn't capture this by default. Decided not to add this for now as the dashboards work without. Can be a future enhancement. <img width="1563" height="487" alt="Screenshot 2026-03-19 at 2 45 18 PM" src="https://github.com/user-attachments/assets/51028e16-ee2c-45a9-9025-26f17b0db67a" /> # Checklist for submitter ## Testing - [x] QA'd all new/changed functionality manually <!-- This is an auto-generated comment: release notes by coderabbit.ai --> ## Summary by CodeRabbit ## Release Notes * New Features * Added a new observability dashboard for database and connection performance metrics, including RPS, latency, connection pool saturation, and queue statistics. * Enhanced database metrics collection with automatic registration of connection and query performance indicators. * Standardized OpenTelemetry metric naming to align with industry conventions for improved observability compatibility. <!-- end of auto-generated comment: release notes by coderabbit.ai -->	2026-03-20 11:07:58 -05:00
..
database_custom_dashboard.json	Added OTEL DB stats metrics, renamed trace attributes to expected OTEL names (#42097 )	2026-03-20 11:07:58 -05:00
README.md	Added contributor docs for SigNoz. (#39402 )	2026-02-09 15:28:28 -06:00

Konstantin Sykulev 6ed3ba6801

Added OTEL DB stats metrics, renamed trace attributes to expected OTEL names (#42097 )

1. Added DB metrics via otelsql.RegisterDBStatsMetrics()
`db.sql.connection.open`
`db.sql.connection.max_open`
`db.sql.connection.wait`
`db.sql.connection.wait_duration`
`db.sql.connection.closed_max_idle`
`db.sql.connection.closed_max_idle_time`
`db.sql.latency.*`
2. renamed these metrics to signoz convention/expected names
`db.sql.connection.open` -> `db.client.connection.usage`
`db.sql.connection.max_open` -> `db.client.connection.max`
`db.sql.connection.wait` -> `db.client.connection.wait_count`
`db.sql.connection.wait_duration` -> `db.client.connection.wait_time`
`db.sql.connection.closed_max_idle` -> `db.client.connection.idle.max`
`db.sql.connection.closed_max_idle_time` ->
`db.client.connection.idle.min`
3. created custom dashboard to display these metrics, (import via json)
<img width="1580" height="906" alt="Screenshot 2026-03-19 at 2 44 43 PM"
src="https://github.com/user-attachments/assets/f1b64ed6-e534-4490-8955-bc1205dd21d4"
/>
4. Fixed metrics for service db dashboards
Signoz expects

`db.system` : Identifies the database type (e.g., postgresql, mysql,
mongodb).
`db.statement` : The actual query being executed (e.g., SELECT * FROM
users).
`db.operation` : The type of operation (e.g., SELECT, INSERT).
`service.name` : The name of the service making the call.

We needed to set the `db.system` attribute explicitly.

`db.operation` is missing because otelsql doesn't capture this by
default. Decided not to add this for now as the dashboards work without.
Can be a future enhancement.

<img width="1563" height="487" alt="Screenshot 2026-03-19 at 2 45 18 PM"
src="https://github.com/user-attachments/assets/51028e16-ee2c-45a9-9025-26f17b0db67a"
/>


# Checklist for submitter

## Testing
- [x] QA'd all new/changed functionality manually

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->

## Summary by CodeRabbit

## Release Notes

* **New Features**
* Added a new observability dashboard for database and connection
performance metrics, including RPS, latency, connection pool saturation,
and queue statistics.
* Enhanced database metrics collection with automatic registration of
connection and query performance indicators.
* Standardized OpenTelemetry metric naming to align with industry
conventions for improved observability compatibility.

<!-- end of auto-generated comment: release notes by coderabbit.ai -->

2026-03-20 11:07:58 -05:00

database_custom_dashboard.json

Added OTEL DB stats metrics, renamed trace attributes to expected OTEL names (#42097 )

2026-03-20 11:07:58 -05:00

README.md

Added contributor docs for SigNoz. (#39402 )

2026-02-09 15:28:28 -06:00

README.md

Running SigNoz locally with Fleet

SigNoz is an open-source observability platform that provides traces, metrics, and logs in a single UI. This guide explains how to run SigNoz locally for Fleet development with optimized settings for reduced latency.

Prerequisites

Docker and Docker Compose
A locally-built Fleet server (see Testing and local development)

Setup

Clone the SigNoz repository at a specific release:

git clone --branch v0.110.1 --depth 1 https://github.com/SigNoz/signoz.git
cd signoz/deploy

Modify the SigNoz UI port to avoid conflict with Fleet (which uses port 8080):

In docker/docker-compose.yaml, change the signoz service port mapping:

services:
  signoz:
    ports:
      - "8085:8080"  # Changed from 8080:8080 to avoid conflict with Fleet

(Optional) For reduced latency during development, modify docker/otel-collector-config.yaml:

processors:
  batch:
    send_batch_size: 10000
    send_batch_max_size: 11000
    timeout: 200ms  # reduced from 10s for dev
  # ...
  signozspanmetrics/delta:
    # ...
    metrics_flush_interval: 5s  # reduced from 60s for dev

Start SigNoz:

cd docker
docker compose up -d

Give it a minute for all services to initialize. The SigNoz UI will be available at http://localhost:8085.

Configuring Fleet

Start the Fleet server with OpenTelemetry tracing and logging enabled:

export FLEET_LOGGING_TRACING_ENABLED=true
export FLEET_LOGGING_OTEL_LOGS_ENABLED=true
export OTEL_SERVICE_NAME=fleet
export OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:4317

./build/fleet serve

Note: All log levels (including debug) are always sent to SigNoz regardless of the --logging_debug flag. That flag only controls stderr output.

Low-latency configuration (optional)

For faster feedback during development, you can reduce the batch processing delays on the Fleet side:

# Batch span processor delay (default 5000ms)
export OTEL_BSP_SCHEDULE_DELAY=1000

# Log batch processor settings
export OTEL_BLRP_EXPORT_TIMEOUT=1000
export OTEL_BLRP_SCHEDULE_DELAY=500
export OTEL_BLRP_MAX_EXPORT_BATCH_SIZE=1

./build/fleet serve

Using SigNoz

After starting Fleet with the above configuration, you should start seeing traces, logs, and metrics in SigNoz UI at http://localhost:8085.