Skip to content
Case study · Interface engines

One console for Mirth, OIE and BridgeLink engines

Integration engines fail quietly: a channel stops, a queue grows, a lab feed goes silent. We built a module that watches three engine flavours, alerts on four signals every two minutes, and knows which Mirth versions a clinic may legally run.

Network patch panel with neatly routed cables in a server room
Case study at a glance
  • 3 enginesOIE, Mirth ≤4.5.2, BridgeLink
  • 4 alert metricsQueue, silence, errors, heap
  • Every 2 minutesAlert rules evaluated
  • Licence-aware4.5.2 ceiling in code

The problem

Clinics and labs often run an integration engine with nobody watching it. Built-in alerts go to a shared mailbox, and the first sign of a stopped channel is a clinician asking where a result went. Since 2025, which Mirth version you run is also a licensing question.

What we built

An Engine module in Radiatus OneHealth: register an engine, let a small agent beside it report channel state and counters, and evaluate alert rules on queue depth, feed silence, error rate and JVM heap every two minutes.

Where it stands

Running inside OneHealth for Open Integration Engine, Mirth Connect up to 4.5.2 and BridgeLink. The same signals are the ones we set up for clients in our interface engine support work.

Numbers you can check

Every figure on this page comes from the running product or its source code, checked on 7 November 2026. None is a client result.

FigureWhat it countsWhere it comes from
3Engine flavours the module accepts: Open Integration Engine, Mirth Connect up to 4.5.2, BridgeLinkThe module's engine registry
4.5.2Highest Mirth Connect version accepted; 4.6 and later are commercially licensed by NextGenA ceiling written into the registry, with the reason in a code comment
4Alert metrics: queue depth, feed silence, error rate, JVM heap percentageThe module's alert rules
2 minHow often alert rules are evaluated for each tenantThe module's scheduled job
5Channel states tracked: running, paused, stopped, error, unknownThe module's database schema
3Counters per channel: received, sent, erroredThe module's database schema

Why engine monitoring needed its own module

An integration engine sits between the EHR, the lab, the scheduler and the payer. When it misbehaves, nothing crashes visibly. Results stop arriving, a queue backs up behind a partner system that is down, or a feed that normally sends a message every few minutes simply goes quiet. Engines have their own alerting, but in small clinics and labs it usually emails an inbox nobody reads, and nobody owns the engine day to day.

OneHealth already connects to a practice's EHR, so watching the engine that feeds it was the natural next step. We wanted signals a non-specialist could act on, sent to the right person, and nothing that required changing how the engine itself was configured.

The four signals, and why these four

  • Queue depth catches a destination that is down or slow before messages pile up for hours.
  • Feed silence catches the failure no error counter shows: an upstream system that stopped sending. It is measured in minutes since the last message on a channel that should be busy.
  • Error rate catches mapping and validation failures, usually after a partner changes their feed without warning.
  • Heap percentage catches a Java engine heading for an out-of-memory restart, typically because message stores are never pruned.

Rules are evaluated every two minutes per tenant. The interval is deliberate: a feed-silence threshold expressed in minutes means nothing if it is checked less often than that. Old metrics are pruned nightly so the monitoring database does not become the next thing that fills a disk.

Licence-aware by design

From version 4.6, NextGen distributes Mirth Connect under a commercial licence. The open-source line continues as community forks such as the Open Integration Engine and BridgeLink. The module's engine registry accepts Mirth Connect only up to 4.5.2, and the reason is written next to the limit in the code: the ceiling is legal, not technical. A clinic that needs a newer Mirth version needs a NextGen licence, and our tooling should not quietly suggest otherwise.

For teams deciding where to go next, our Mirth Connect migration guide covers the options and the work involved.

Security choices

Each registered engine gets an ingest secret that is shown once and stored only as a hash, so it cannot be read back from the screen or the database. The agent beside the engine posts metrics outward; the module never needs inbound access to the clinic's network. Engines, channels and alerts are scoped to a tenant like every other OneHealth record.

The stack

LayerChoiceWhy
ModuleOneHealth Engine module, PHP 8.3 and MariaDB 10.11Shares the platform kernel, tenancy and audit log
CollectionA lightweight agent beside the engine posting metrics over HTTPSNo inbound firewall rules at the clinic
EnginesOpen Integration Engine, Mirth Connect up to 4.5.2, BridgeLinkThe open-source engines clinics actually run
SchedulingDatabase job queue: alerts every 2 minutes, pruning nightlyNo extra infrastructure to run
SecretsPer-engine ingest secret, stored hashedA leaked screen or database row reveals nothing usable

What it does not do yet

  • It watches and alerts. It does not restart channels, reprocess messages or deploy channel changes.
  • Metrics come from an agent installed beside the engine; without it the module has nothing to evaluate.
  • Commercial engines such as Rhapsody, Iguana or Mirth Connect 4.6 and later are not supported by this module.

What this means for your interface engine

The same four signals are the starting point of every interface engine support engagement we run, whether the engine is Mirth Connect, the Open Integration Engine or something else. We add them to your existing monitoring, route alerts to people who can act, and write the reprocessing runbooks that go with them.

If your engine is old, undocumented or stuck on a version you are unsure you may run, start with our Mirth Connect services review. For a wider view of engine choices, read Mirth vs Rhapsody vs Iguana.

FAQ

Questions about this case study

Open Integration Engine, Mirth Connect up to version 4.5.2 and BridgeLink. Mirth Connect 4.6 and later are commercially licensed by NextGen and are not accepted by the module.

Four metrics: queue depth, feed silence (minutes since a channel last received a message), error rate and JVM heap percentage. Rules are evaluated every two minutes.

No inbound access. A small agent beside the engine posts metrics outward over HTTPS using a per-engine secret that is stored only as a hash.

No. It detects and alerts. Restarting channels and reprocessing messages are done by people following runbooks, which is part of our interface engine support service.

Yes. We add channel, queue, silence and error monitoring to Mirth, Open Integration Engine and other engines, route alerts to your on-call people and document the response for each alert.

Work with the team that built it

Need something like this for your organisation?

Tell us what you are connecting, running or securing. The engineers behind this case study reply personally.

  • A senior engineer reads every enquiry
  • Happy to sign an NDA first
  • You get a written next step, not a sales sequence
Last updated