Backend Logs

The Logs page streams real-time application logs from the Rackscope backend directly in the Web UI, without needing SSH or docker logs.

Accessing Logs

Navigate to Logs in the bottom of the sidebar, or go to /logs.

The link is visible by default. To hide it (kiosk or public deployments):

# config/app.yaml
features:
  show_logs: false

Or toggle it in Settings → Views → Pages & Navigation → Backend Logs.

---

Interface

Live mode vs Paused mode

Mode	Behaviour
Live (default)	Connects via SSE — new log lines appear in real time (~0.5 s latency). Green pulsing dot in the button.
Paused	Disconnects SSE. Shows a snapshot. Use Refresh to manually reload.

Click the Live / Paused button in the page header to toggle.

Filters

Control	Description
Level pills	Filter by ALL, DEBUG, INFO, WARNING, ERROR.
Search bar	Case-insensitive substring match on message and logger name.
Entry count	Shows the number of lines currently displayed.

Reverse order

The ↕ icon button toggles between:

• Oldest first (default) — chronological, auto-scroll follows the bottom
• Newest first — most recent entry at the top, auto-scroll follows the top

Auto-scroll is disabled when you scroll away from the tracked edge and re-enables when you scroll back.

Actions

Button	Action
Live / Paused	Toggle SSE stream
Refresh (paused only)	Reload snapshot
↕	Toggle reverse order
↓ Export	Download current buffer as rackscope-logs-<timestamp>.json
Clear	Empty the server-side buffer immediately

Row detail

Click any log row that has an exception or a request_id to expand it:

• request_id — correlates with the X-Request-ID response header
• Exception — full Python traceback, pre-formatted

---

Technical details

In-memory ring buffer

Logs are kept in a 1 000-entry ring buffer in RAM. The buffer is:

• Never persisted to disk
• Cleared when the backend container restarts
• Cleared via the Clear button or DELETE /api/logs

For persistent log storage, mount a volume and redirect stdout to a file, or use a log aggregator (Loki, Elastic, etc.).

Sensitive field redaction

Secrets are stripped at capture time, before entering the buffer:

Pattern	Redacted to
URL credentials http://user:secret@host	http://user:***@host
password=…, token=…, secret_key=…, etc.	password=***

SSE stream limits

Parameter	Value
Max concurrent streams	5
Auto-disconnect after	10 minutes
Heartbeat interval	30 seconds
Poll interval	0.5 seconds

API endpoints

All endpoints require admin access (require_admin dependency — see Admin Endpoint Protection).

Method	Endpoint	Description
GET	/api/logs	Recent entries. Params: n (max 1000), level, search, since_seq
GET	/api/logs/stream	SSE stream. Params: level, search
DELETE	/api/logs	Clear the buffer

---

Log levels reference

Level	Colour	Meaning
DEBUG	Gray	Verbose internal state — disabled by default
INFO	Blue	Normal operations: requests, topology reload, cache hits
WARNING	Amber	Non-critical issues: missed cache, slow query, retried request
ERROR	Red	Failures requiring attention: Prometheus unreachable, YAML parse error
CRITICAL	Red bold	Fatal errors — backend may be in a degraded state

Default log level is INFO. Change it with the environment variable:

RACKSCOPE_LOG_LEVEL=DEBUG