Skip to main content

Set up Lattice in an environment

Outcome. Lattice is installed and licensed, people hold the roles they need, the background processes run, your logging, guardrail and event choices are made, and a test call shows up where operators look. Who takes part: Integration administrator, Data management consultant, Integration developer, Integration operator. Stages that use standard pages (Assign users to roles, Microsoft Entra ID applications, Process automations, the standard Business events catalog, the Siren pages) need System administrator or the access named on the role page; the Lattice roles do not open them. How often: once per environment. To move an environment that already runs Lattice to a new release, follow Take a Lattice or Finance and Operations update.

1. Install the package (Integration administrator)​

One package carries Lattice and its licensing module, so this stage is a single deployment followed by a check that the environment accepts the license.

Done when: all 13 Lattice service groups are listed at https://<your-environment>/api/services (the list), and the Lattice line on Siren license diagnostics does not read Expired, Invalid or Offline exceeded.

2. Give people their roles (System administrator)​

Decide who operates, who only reads, and who may run the high-risk and recovery actions, before anyone starts work. Integration clients get their own identities and roles later, in Onboard a new integration.

Done when: on Assign users to roles, your operators hold Integration operator, reviewers and auditors hold Integration read-only, at least one person you trust holds Integration administrator (the refreshes in the next stage and the recovery actions need it), and Database access administrator has no users or only named people.

3. Check the background work and fill the catalogs (Integration administrator, System administrator)​

Retries, reconciliation, statistics and log cleanup run as background processes, which the deployment's database synchronization registers, so check that they run. The explorers and the event catalog read stored catalogs that nothing fills by itself, and until the service metadata refresh has run, the Service test console cannot run an operation.

Done when: the Background processes tab of Process automations lists the Lattice ... processes, including Lattice Async Operation Processing; the Service metadata explorer tree shows the Lattice service groups; the Catalog status line of the OData entity metadata explorer no longer reads Not yet refreshed; and Last refreshed on the Business event catalog is later than the deployment.

4. Choose how Lattice behaves (Integration administrator)​

Each of these settings starts off, empty or at a working default, so set only what your business needs. Decide before the first integration goes live, because they shape what the logs keep and which orders are held.

Done when: Service operation logging configuration holds a row only for each operation you chose to log less or keep for a different time, and, if you turned the inbound order policy on, Inbound order policy evaluation runs as a recurring batch job and new orders on Inbound order policy state move out of Pending.

5. Switch on alerts and business events (System administrator, Integration administrator)​

An event is sent only when it is activated in the standard business events catalog, which the Lattice roles do not open, so someone with access to it does this. Activate the alert events now, so the people who monitor are told when something needs them; the events an integration consumes are activated when it is onboarded.

Done when: Integration item parked, Business event publish failed and Workflow stalled or failed, and Recurring data job health changed if you use recurring data jobs, are active in the standard Business events catalog with your endpoint and the legal entities you chose.

6. Prove it end to end (System administrator, Integration developer, Integration operator)​

A smoke test proves the pieces work together: a test client makes a read-only call and a preview, and an operator finds the preview where they will look for failures later.

Done when: both calls return Success true, the preview is on Service call log with Preview set (a status query is never logged), and the Integration operations workspace opens from the default dashboard with its nine count tiles.

If something goes wrong​