Monitoring Dashboard¶

LogEverything includes a web-based monitoring dashboard for exploring logs, operations, and system metrics collected by the monitoring subsystem.

Getting Started¶

Prerequisites: Python 3.8+ and a data directory containing the SQLite database (monitoring.db) and/or JSONL log files produced by LogEverything’s monitoring subsystem.

cd logeverything-dashboard
pip install -r requirements.txt
python run_dashboard.py
# Dashboard opens at http://localhost:3001

The startup script compiles SCSS, launches a file-watcher for live style reloading, and starts the Uvicorn server.

Connecting to Data¶

On first launch a Default Connection pointing at ../sample_monitoring_data is created. To connect to your own data:

Click the link icon in the sidebar footer.
Choose Local (file path) or API (remote endpoint).
Provide the path to the directory containing monitoring.db and/or *.jsonl log files.
Click Test Connection, then Save.

Note

Local connections automatically discover both active log files (*.jsonl) and rotated files (*.jsonl.*, including gzipped archives). No extra configuration is needed — point the connection at your data directory and all matching files are read.

Layout¶

The dashboard uses a multi-page layout with a fixed sidebar and a header bar.

Sidebar (left, 240 px) — Navigation links, connection status, version. The active page is highlighted. Collapses on viewports narrower than 1024 px and can be toggled with the hamburger icon.
Header (top) — Time-range selector, auto-refresh toggle, export button, and theme toggle.

Pages¶

Overview¶

The landing page gives a high-level snapshot:

Summary cards — System Stats, Monitoring, Logs, Operations (auto-refresh every 10 s).
System Trends — CPU and Memory time-series chart.
Log Distribution — Doughnut chart by log level.
Operation Analytics — Top operations with counts and failure rates.

Logs¶

A dedicated page for exploring log entries.

Stats card — Total logs, error count, active sources.
Log Distribution chart — Doughnut chart by level.
Filter bar — Level pills (multi-select), source, correlation ID, full-text message search. Filters are applied with the Apply button and persist across auto-refresh cycles.
Pagination — Choose page size (5 / 25 / 50 / 100 / ALL) or enter a custom value. The footer shows “Showing X–Y of Z” with Prev / Next navigation.
View toggle — Switch between Flat (table) and Tree (hierarchical) views.

Operations¶

Summary cards — Total operations, failure rate, avg/max duration, success rate.
Operation Analytics — Top operations table with failure rates.
Recent Operations — Table with status badges, duration, errors, and timestamps.

System¶

System Stats card — CPU, Memory, Disk, Network.
System Trends chart — Full-width chart for detailed trend analysis.
System Detail — Process metrics (RSS, VMS, threads, CPU, file descriptors) and resource metrics (swap, disk I/O, network I/O).
Session info — Session ID, start time, working directory, uptime.

Tree View¶

When logs are emitted through LogEverything’s @log decorator with a JSONHandler or JSONLineFormatter, each record carries structured hierarchy fields (indent_level, call_id, parent_call_id, log_type, execution_id). The dashboard reconstructs a collapsible tree from these fields.

JSONLineFormatter can be attached to any handler — including the rotation handlers — making it the recommended choice when you need both file rotation and dashboard integration. See File Rotation for examples.

On the Logs page, click Tree to switch views.

Expand / Collapse — Click a function call header to toggle children.
Duration badges — Elapsed time from paired call_entry / call_exit records.
Level badges — Colour-coded log level on each node.
Filter-aware — Tree view respects all active filters (level, source, correlation ID, search, time range).

Note

The tree view requires hierarchy fields in the log data. Logs collected without @log or HierarchyFilter will show a flat list. See Structured Hierarchy Fields for details.

Filtering & Pagination¶

Level pills — Click one or more level badges (DEBUG, INFO, WARNING, ERROR, CRITICAL) to toggle them. Active pills are filled with the level colour. Multiple levels can be selected simultaneously.

Text filters — Source, Correlation ID, and Search fields narrow results further. Press / to focus the search input.

Time range — The header buttons (1 h / 6 h / 24 h / 7 d) filter all pages. The default is 24 h.

Pagination — The ROWS selector above the table lets you choose 5, 25, 50, 100, ALL, or a custom page size. The table footer shows the current range and Prev / Next buttons.

All filters and pagination settings persist across the 10-second auto-refresh cycle. Changing a filter resets to page 1. Filters also apply when the Tree view is active (pagination is hidden in tree mode).

Themes¶

Click the sun/moon icon to toggle between dark and light themes.

The preference is saved in localStorage. Charts are recreated on theme change to pick up the new colour tokens.

Keyboard Shortcuts¶

Key	Action
`r`	Toggle auto-refresh on/off
`/`	Focus the log search input
`e`	Export data as JSON
`Escape`	Close any open modal

Auto-Refresh & Export¶

HTMX partials poll every 10 seconds. Charts refresh every 30 seconds. Click the refresh icon (or press r) to pause all polling; click again to resume.

Click the download icon (or press e) to export a JSON file containing system stats, metrics history, operations, logs, and session info for the selected time range.

API Reference¶

Page Routes¶

Route	Description
`GET /`	Overview
`GET /logs`	Logs
`GET /operations`	Operations
`GET /system`	System

HTMX Partials¶

Route	Parameters
`/partials/system-stats`
`/partials/monitoring-stats`
`/partials/logs-stats`
`/partials/operations-summary-card`
`/partials/operations`	`?hours=` `?operation_name=`
`/partials/logs`	`?level=` `?source=` `?q=` `?hours=` `?limit=` `?offset=`
`/partials/logs-tree`	`?level=` `?source=` `?q=` `?hours=` `?correlation_id=`
`/partials/operation-analytics`	`?hours=`
`/partials/system-detail`
`/partials/trace/{correlation_id}`

JSON APIs¶

Route	Description
`/api/system-metrics-history`	Time-series data (`?hours=` `?max_points=`)
`/api/system-detail`	Latest system snapshot
`/api/operations/summary`	Aggregated analytics (`?hours=`)
`/api/log-level-distribution`	Log count by level (`?hours=`)
`/api/logs/tree`	Hierarchical tree JSON (`?hours=` `?limit=`)
`/api/session-info`	Current session metadata
`/api/export`	Full data export download (`?hours=`)
`POST /api/ingest/logs`	Receive batched logs from remote transports

Customisation¶

The dashboard uses SCSS compiled to main.css. Source files are in dashboard/static/scss/:

_variables.scss — Colours, fonts, spacing, radii, transitions.
components/ — One file per UI component (sidebar, tree, pagination, etc.).
themes/ — _dark.scss and _light.scss override CSS custom properties.

To recompile after edits:

python -m dashboard.compile_scss          # one-shot
python -m dashboard.compile_scss --watch   # continuous

The run_dashboard.py script compiles and watches automatically.