Settings

Observability settings

Global trace-capture mode plus the per-tenant sync-failure alert thresholds the scheduled watcher pages on. Applies across the whole deployment; super-admin applies changes directly.

Trace capture mode

observability_settings.trace_capture_mode

How much request/job tracing is retained. Higher modes raise trace volume and cost — flip to Full only for a debugging session, then drop back.

Off
No traces retained. Metrics and logs are unaffected. Errors only
Keep traces for failed and errored requests/jobs; drop the rest. Sampled default
Tail-keep: 100% of errors + slow traces, ~10% of successes. Full high volume
Capture-all: 100% retention. Use briefly for live debugging.

Sync-failure alert — global default

observability_settings.sync_alert_*

Alert when failure rate exceeds

% of syncs

Over a window of

Notify via

Email Slack

Evaluated per tenant on an hourly cadence. A tenant over the threshold is paged once per window. Recipients are super-admins only.

Per-tenant thresholds

Effective value resolves to the global default unless a per-tenant override is set (overrides deferred).

Tenant	Threshold %	Effective value
Acme Ads Co.	— inherit —	20% over 1 day default
Bharat Media	— inherit —	20% over 1 day default
Chai Labs	— inherit —	20% over 1 day default

v1 ships one global threshold; the table shows the resolved effective value per tenant. A future packet adds editable per-tenant overrides.