REST API

Authenticated routes live under {path}/api/* (default /loadflux/api). The dashboard UI uses these; you can call them from scripts or external tools with the same session cookie or Bearer token as login.

Unless noted, endpoints require auth.

Common query parameters

Parameter	Used on	Description
from, to	Most routes	Unix timestamps in milliseconds (inclusive range)
page, limit	List routes	Pagination; page ≥ 1, limit 1–1000 (default 200)
max_points	/system, /process	Server-side downsampling cap (10–2000). If omitted and range ≥ 5 days, the server auto-samples to 900 points
instance	/system, /process	Filter to one instance_id. When cluster mode is on and omitted, responses use cluster aggregate
search	Endpoints, errors, slow	Case-insensitive substring filter
status	Errors	Filter error log to one HTTP status code
metric, limit	/endpoints/top	Top-endpoint sort key and count (default metric request_count, limit 10)
threshold	/endpoints/slow	Slow-request threshold in ms (defaults to configured slowRequestThreshold)

Routes

Live

Method	Path	Description
GET	/sse	Server-Sent Events stream (DashboardSnapshot every ~2s). Max 64 concurrent clients per process
GET	/snapshot	One-shot JSON snapshot (same shape as SSE)

Cluster

Method	Path	Description
GET	/instances	{ instances: string[] } — distinct instance_id in from/to

Metrics

Method	Path	Description
GET	/system	System time series (SystemMetricRow[] or paginated)
GET	/process	Process time series (ProcessMetricRow[] or paginated)
GET	/endpoints	Endpoint aggregation rows
GET	/endpoints/top	Top endpoints by metric
GET	/endpoints/slow	Slow endpoint windows
GET	/errors	Error log rows
GET	/errors/status-codes	{ codes: number[] } for the range
GET	/errors/distribution	Status bucket totals (StatusDistribution)
GET	/overview	Aggregated overview for the range (OverviewMetrics)

Settings and export

Method	Path	Description
GET	/settings	{ retention_days, slow_threshold }
POST	/settings	JSON body: optional retention_days (≥ 1), slow_threshold (≥ 0)
GET	/export	JSON bundle { system, process, endpoints, errors }; max range 31 days. In cluster mode, system/process series are cluster-aggregated

Auth (unauthenticated where noted)

Login and setup routes are documented in Authentication. Initial setup uses POST /auth/setup when no user exists.

Example

# System metrics for the last hour (cluster aggregate when cluster mode enabled)
curl -s -b cookies.txt \
  "http://localhost:3000/loadflux/api/system?from=$(($(date +%s)*1000-3600000))&to=$(($(date +%s)*1000))&max_points=280"

# One replica only
curl -s -b cookies.txt \
  "http://localhost:3000/loadflux/api/system?from=0&to=9999999999999&instance=apprunner-task-a"