Monitoring & Integrations
MTE Relay v5 is built to be observed through its structured JSON logs. Every request produces correlated, machine-parseable events — performance timings (including the MTE decrypt/validate/proxy/encrypt phase breakdown), security events (signature failures, replay signals, state-store health), encryption audit events (hashes only, never plaintext), and error events — all keyed by a shared correlation_id and tagged with service, product (MRS/MAR), relay_version, and instance.
No vendor SDK runs inside the relay. Your log collector ships stderr to your platform of choice, and the prebuilt assets on this site do the rest.
Start here
- Event Schema Reference — the field-level contract all integrations build on.
- Pick your platform:
| Platform | What's provided |
|---|---|
| Datadog | Log pipeline, facet/measure list, importable dashboard, five monitors |
| New Relic | Importable two-page dashboard, alert policy with five NRQL conditions |
| Grafana | Dashboards for CloudWatch Logs Insights (AWS) and Loki (Kubernetes/on-prem) |
Splunk, Elastic, and any other JSON-capable aggregator work the same way — ingest the JSON lines and build on the schema reference.
Recommended alerts (any platform)
- Any
security_eventwith severityhigh/critical - Any
error_event.event_type = panic - Sustained
state_save_failed/state_not_found(state store health) - Average
performance_event.total_time_msabove your latency budget - Spikes in
client_pairs_exhausted(capacity early-warning)
Cost control
Per-request performance events dominate log volume. High-traffic deployments should set PERF_EVENT_SAMPLE_PERCENT (e.g. 10) — dashboard rates and percentiles stay statistically sound, and security/error/audit events are never sampled.