Skip to main content

Configuration reference

Minimal config

BridgeConfig.yaml
bridge:
facility-code: BRANCH-01
hapi-fhir-url: http://hapi:8080/fhir
consent:
enabled: true
fail-closed: true

Full reference

BridgeConfig.yaml
bridge:
facility-code: BRANCH-01
facility-name: Apollo Branch 01

hapi-fhir-url: http://hapi:8080/fhir
hapi-fhir-timeout-ms: 5000
hapi-fhir-circuit-breaker:
enabled: true
failure-threshold: 5
wait-duration-s: 30
retry-attempts: 10

mpi:
url: http://jammpi:8081/fhir
cache-ttl-minutes: 5
min-grade: PROBABLE

facility-registry:
url: http://jamfr:8089/fhir
cache-ttl-minutes: 5
auto-provision: true

consent:
enabled: true
fail-closed: true
url: http://jamconnect:8091/jamconnect/consent/check
timeout-ms: 3000

terminology:
enabled: true
url: http://jamts:8085/fhir
translate-local-loinc: true

deduplication:
enabled: true
window-hours: 24

mllp:
adt: 2575
oru: 2576
orm: 2577
mdm: 2578
qbp: 2579
rde: 2580
siu: 2581
mfn: 2582

subscriptions:
enabled: true
seed-default: true

audit:
enabled: true
transport: both
syslog-host: auditstore
syslog-port: 6514
fhir-balp-url: http://hapi:8080/fhir

dhis2:
enabled: false
url: http://dhis2/api
username: admin
password: ${DHIS2_PASSWORD}
notifiable-loincs:
- "80823-7"
- "32700-2"
- "11545-1"
- "625-4"

Key decisions

Always use fail-closed in production

consent.fail-closed: false means an unreachable consent service is treated as a permit. This is not a safe default for clinical data. Only use false in local development where the consent service may not be running.

consent.fail-closed

When true, JamBridge NACKs if the consent service is unreachable. Recommended for production. When false, an unreachable consent service is treated as permitted — do not use in production.

mpi.min-grade

GradeScoreUse case
CERTAIN≥ 0.95High national ID coverage
PROBABLE≥ 0.85Mixed ID coverage (recommended)
POSSIBLE≥ 0.70Low ID coverage, manual review

hapi-fhir-circuit-breaker

When the circuit opens (5 consecutive failures), messages queue in SecondaryRetryQueue with exponential backoff. The sending HIS receives MSA|AA — it is not aware the FHIR write is pending. The queue drains automatically when HAPI recovers.