Operational Metrics & Capacity
A runbook for monitoring request volume, verdict distributions, and data retention.
The NjiraAI Console provides top-down visibility into your gateway activity and enforcement metrics. This guide details how to use the Usage dashboard to monitor the health of your agent fleet.
Operational Monitoring Checklist
To access your telemetry, navigate to the Usage tab in the Console. During large rule rollouts or major agent releases, monitor the following charts daily:
1. Total Request Volume
- What it tracks: Aggregate requests processed by the NjiraAI Gateway or deep SDK integrations.
- What to look for: Sudden spikes (which might indicate a looping agent or a DDoS attack) or sudden drops (which might indicate a broken client integration or severe upstream outage).
2. Verdict Distribution
- What it tracks: The ratio of
ALLOW,BLOCK,MODIFY, andREQUIRE_APPROVALverdicts over time. - What to look for: A baseline shift in
BLOCKvolume. If blocks suddenly spike 5x, your agent might be trapped in a failure loop attempting an illegal action repeatedly, or a recent policy change was too aggressive (a false positive wave).
3. Top Policy Hits
- What it tracks: A leaderboard of which specific policy rules trigger most frequently.
- What to look for: Identify excessively noisy rules. If a
shadowmode rule is dominating the chart, it urgently requires tightening via detectors or scope before promotion toactive.
Capacity and Limits (Free Tier)
The default free tier provides generous capacity for development and small-scale production deployments. Usage limits and your current consumption burndown are displayed continuously on the Usage page.
Included in default capacity:
- Full enforce latency and bandwidth for all built-in policy packs.
- Unlimited Gateway proxy connections.
- Console access with trace viewer, playback, and policy authoring.
- 30-day trace telemetry retention.
If your consumption bar approaches the limit, contact the NjiraAI team to configure enterprise throughput limits.
Data Retention and Pruning
NjiraAI retains audit logs and traces based on your organization's compliance configuration. Traces older than the retention period are automatically and permanently pruned via an asynchronous background job.
| Retention Tier | Duration |
|---|---|
| Default | 30 days |
| Enterprise | Configurable (90 days, 1 year, indefinite) |
Retrieving Historical Data
If you require long-term cold storage of your audit logs beyond your active retention window, you must periodically export your trace data via the API and ingest it into your internal SIEM or cold storage buckets.