Logging and audit
Collect structured Log4j2 events and use the transfer journal and signed evidence to investigate AS4 deliveries.
Use operational logs to diagnose a service condition, the transport journal to reconstruct individual attempts, and signed evidence to inspect a specific receipt. Each source has a different access scope.
Collect operational events
Section titled “Collect operational events”The bundled Log4j2 configuration writes one JSON object per line to standard output. Configure the service host or container log driver to collect and retain that stream. For a Docker container, inspect recent output with:
docker logs --since 15m <access-point-container>| Field | What it contains |
|---|---|
schema_version, timestamp |
Format version and event time in UTC. |
service, engine_version |
Runtime and phase4 engine identifiers. |
stage, outcome, reason |
Fixed event stage, result and reason code. |
operation_id, correlation_sha256 |
Operation UUID and an optional hash of the internal correlation value. |
duration_ms |
Elapsed time in milliseconds, when measured. |
Group events by stage and outcome, then use the bounded reason code to investigate. The event format excludes package bytes, credentials, certificate material and raw exception text. Keep wire and TLS debug logging out of the normal operating profile. Repeated database-readiness and recovery failures are rate-limited in the operational event stream.
For example, a database probe failure has the following shape:
{ "schema_version": 1, "timestamp": "2026-01-01T10:00:00Z", "service": "sdk-access-point-phase4", "engine_version": "4.3.0", "stage": "DATABASE", "outcome": "TEMPORARY_FAILURE", "operation_id": "502612b8-57d4-41bc-b4b6-4e9dae1ea996", "correlation_sha256": "", "reason": "DATABASE_UNAVAILABLE", "duration_ms": 0}Set rotation and retention in the host’s log collector. Keep separate retention and access policies for routine logs, journal extracts, signed evidence and stored package data.
Read the transport journal and receipt evidence
Section titled “Read the transport journal and receipt evidence”| Source | Use and access |
|---|---|
phase4_dispatch |
Current outbound transport state and attempts. Use the private status API for application lookups. |
phase4_transfer_journal |
Append-only, tenant-scoped history including attempt and dispatch identifiers, AS4 message metadata, outcome, reason and evidence reference. Journal extraction uses audit:read. |
phase4_signal_evidence |
Signed AS4 signal, signing certificate and SHA-256 digest. Evidence retrieval uses audit:evidence:read. |
Journal and evidence extraction are protected runtime operations for an authorized operator integration. Scope extracts to the affected tenant and UTC interval. The journal API accepts a limit up to 10,000 rows and rejects results that exceed the requested limit. Use shorter intervals for larger investigations.
ATTEMPT_STARTED records intent before network I/O. Look for the matching ATTEMPT_OUTCOME and signed evidence before deciding what the peer received. On the receiving side, SIGNAL_SENT with SIGNED_RECEIPT_PREPARED records that the receipt was prepared for the HTTP response. Use the sending access point’s status to establish whether it received and accepted that receipt.
OpenTelemetry
Section titled “OpenTelemetry”When a provider is registered, the runtime emits the counter sdk.ap.events.total with stage and outcome labels, spans named sdk.ap.<stage>, and the log event sdk.ap.event. Configure the provider and exporter for the runtime environment. Keep stdout collection and PostgreSQL evidence in place independently of telemetry export. Monitoring and troubleshooting