Dashboard
The Dashboard is the landing page after login. It rolls up the most useful signals across all projects into one view: live throughput, error rate, latency, top endpoints, recent activity.
What’s on it
Section titled “What’s on it”Hero strip
Section titled “Hero strip”A live-updating row of large indicators:
- Requests / sec — instantaneous throughput (last 60s window).
- Error rate —
errors / totalover the last 60s. Coloured red if > 1%. - p95 latency — 95th percentile latency (ok requests only) over the last 60s.
The strip refreshes every 5 seconds via a small JSON poll.
KPI cards
Section titled “KPI cards”Four cards summarising the past 24 hours:
| KPI | What it means |
|---|---|
| Requests (24h) | Total gateway_logs rows in the last 24h |
| Tokens (24h) | Sum of total_tokens |
| Avg Latency | Mean of latency_ms for status=ok rows |
| Error Rate | count(error) / count(*) as a percentage |
Traffic chart
Section titled “Traffic chart”A stacked area chart of requests per hour over the past 7 days, segmented by provider_key. Lets you spot:
- Hourly cycles (morning peak, evening drop).
- Provider mix shifts (suddenly gpt-4o-mini-heavy after a config change).
- Outage signatures (cliff drops).
Provider status panel
Section titled “Provider status panel”One row per registered provider:
- Status dot (ok / warn / err).
- Total requests / errors in the last 24h.
- Average latency.
- Active credentials count.
Same as the Providers admin page but condensed.
Top endpoints
Section titled “Top endpoints”Top 5 endpoints by request count over the last 24h. Each row links to the endpoint detail page.
Recent activity
Section titled “Recent activity”Last 10 entries from audit_logs. Time-stamped, severity-coloured. Click to filter the audit log by that event.
Refresh + cost
Section titled “Refresh + cost”Most of the data comes from one query against gateway_logs and audit_logs, run on page load. The hero strip uses a separate small endpoint that returns just the live rates.
For high-volume gateways, consider:
- Indexing
gateway_logs(created_at, project_id, status)if it’s not already. - Pruning old rows (
gateway_logs > 90d) to keep queries fast.
Project scope
Section titled “Project scope”The Dashboard is gateway-wide — it sums across every project. For project-scoped views, use Metrics inside the project.
When the gateway is fresh
Section titled “When the gateway is fresh”A brand-new gateway with no traffic shows zero everywhere. That’s expected. Walk through the Quick Start to send your first request — the dashboard updates immediately.
Next: Metrics — project-scoped charts.
© Akyros Labs LLC. All rights reserved.