mherman22

Patient flags: eight issues found migrating an RHD registry onto OpenMRS 3 (patientflags, esm-patient-flags-app, fhir2) - JIRA ticket drafts

Patient flags: nine issues found while migrating a registry onto OpenMRS 3

Context for all eight. We are migrating an RHD (rheumatic heart disease) registry off a bespoke Python system onto OpenMRS 3, and replicating its "critical data flags" with the patientflags module. The old system raised a flag when a field that should have been completed on a form was still empty, and worked them off a queue showing how many days each flag had been pending.

Everything below was reproduced on a running distribution: OpenMRS platform 2.8.10, patientflags

mherman22

How a patient gets saved and sent to OpenCR (SEDISH / iSantePlus)

TL;DR: PatientService.savePatient() → mpi-client AOP advice (PatientSynchronizationAdvice) → async PatientUpdateWorker → MpiClientService.exportPatient() → FHIR conditional-update (upsert) POST to OpenCR on the SEDISH source-key.

Repo (canonical): IsantePlus/openmrs-module-mpi-client (main). Registration trigger lives in charess-org/iSantePlus (registration module).

---

1. The patient is saved — registration module

registration/api/src/main/java/org/openmrs/module/registration/api/impl/RegistrationServiceImpl.java (~line 729)

mherman22

#!/usr/bin/env bash
# xds-sender 2.6.2 activator throws when xdssender.notificationsPullPoint.endpoint is blank,
# which stops xds-sender -> cascades -> isanteplus 'appDashboard' view missing.
# Fix: set the notifications-pull-point globals (non-blank), then restart with a clean cache.
# Values below are placeholders pointing at the SEDISH OpenHIM; change endpoint/creds to the
# real XDS notifications pull point if you actually use that feature.
set -euo pipefail

OMRS=""
for d in /usr/share/tomcat7/.OpenMRS /var/lib/tomcat7/.OpenMRS /root/.OpenMRS /home/*/.OpenMRS; do

mherman22

#!/usr/bin/env bash
set +e
DIR=/usr/share/tomcat7/.OpenMRS
LOG="$DIR/openmrs.log"; [ -f "$LOG" ] || LOG=./openmrs.log

echo "==================== deployed core omods ===================="
ls -la "$DIR/modules/" 2>/dev/null | grep -iE "isanteplus-|coreapps-|registrationapp-|registrationcore-|xds-sender-|labintegration-|m2sys" 
echo

# start of the LATEST boot = last time the scheduler/refresh kicked; use last 'Setting up Country' or last banner.

mherman22

#!/usr/bin/env bash
# Isolates the ROOT fatal error of the most recent OpenMRS boot (the one that
# cancelled the context), rather than errors from earlier boots.
set +e
LOG=/usr/share/tomcat7/.OpenMRS/openmrs.log
[ -f "$LOG" ] || LOG=./openmrs.log
echo "log: $LOG"; echo

echo "-------- last 3 context-cancel / fatal markers (with line#) --------"
grep -nE "cancelling refresh attempt|Unable to start OpenMRS. Error thrown was|Got exception while starting up" "$LOG" | tail -3

mherman22

#!/usr/bin/env bash
# Run in /usr/share/tomcat7/.OpenMRS (or anywhere; it locates openmrs.log).
# Prints the REAL module startup failures + root causes behind the
# "viewProvider isanteplus does not have a view named appDashboard" symptom.
set +e

LOG=""
for f in /usr/share/tomcat7/.OpenMRS/openmrs.log ./openmrs.log /var/lib/tomcat7/.OpenMRS/openmrs.log /root/.OpenMRS/openmrs.log /home/*/.OpenMRS/openmrs.log; do
  [ -f "$f" ] && LOG="$f" && break
done

mherman22

#!/usr/bin/env bash
# SEDISH / server 44: fix the "Continuité des Soins" widget in coreapps patient.gsp.
#  - remove the broken "Check CCD" button (checkCCD() had window.location.href="" -> reload/no-op)
#  - keep ONE button -> registrationResult.page, switched from numeric id to patient uuid,
#    relabelled "Voir le dossier (SEDISH)".
# Edits patient.gsp INSIDE the coreapps .omod (lib-cache is re-exploded from the omod on restart),
# backs up the omod, verifies the edits applied, then restarts tomcat + clears the cache.
set -euo pipefail

OMRS=""

mherman22

#!/usr/bin/env bash
# Locate any xds-sender module copy on this node (to recover the deleted 2.5.8 omod).
set +e
echo "==================== hunting for xds-sender module copies ===================="

OMRS=""
for d in /usr/share/tomcat7/.OpenMRS /var/lib/tomcat7/.OpenMRS /root/.OpenMRS /home/*/.OpenMRS; do
  [ -d "$d/modules" ] && OMRS="$d" && break
done
echo "OpenMRS dir: ${OMRS:-NOT FOUND}"

mherman22

#!/usr/bin/env bash
# SEDISH / server 44: undo the accidental delete of fhir_patient_identifier_system
# rows 3,4 (the previous fix script deleted them, then aborted on a wrong changelog
# table name before it could restore/restart). fhir2 is already started, so its
# liquibase changesets 2026Jul07-1118/-1119 are already recorded -> a restart will
# NOT re-insert 3,4, so we simply restore them from the backup the prior run made.
set -euo pipefail

echo "==================== restore fhir_patient_identifier_system 3,4 ===================="
OMRS=""

mherman22

#!/usr/bin/env bash
# SEDISH / server 44 recovery:
# fhir2 fails to start because its liquibase seed changesets 2026Jul07-1118/-1119
# try to INSERT fhir_patient_identifier_system rows with PK 3 and 4 that already
# exist -> "Duplicate entry for key 'PRIMARY'" -> fhir2 aborts -> 9-module cascade.
#
# Fix: back up the table, drop the two colliding rows AND their (never-recorded)
# changelog entries, then restart so fhir2 re-seeds them canonically.
# Safe & reversible: a full copy of the table is kept as *_bak_recover.
set -euo pipefail