[SFS]-Overlapping Service Appointments Caused by Concurrent Optimization Requests on Territories Sharing Service Resources

Scenario:

Two ORs are triggered at the same time for two different Service Territories. The Service Resources in these territories have the following membership:

Both ORs start simultaneously:

Root Cause

The FSL optimization engine uses an optimistic concurrency model to process scheduled runs:

1. Data Snapshot: At the start of execution, each OR captures a point-in-time snapshot of all relevant resource availability and appointment data.

2. Isolated Execution: The optimization engine runs entirely against this frozen, isolated snapshot.

3. Commit Phase: Upon completion, the OR commits its scheduling results to the database without re-checking whether the underlying data changed during its execution window.

When OR 1 and OR 2 launch simultaneously, they capture an identical data snapshot. The first OR to finish (OR 1) successfully commits its valid assignments. However, when the second OR (OR 2) finishes, it writes changes based on its now-stale snapshot, inadvertently overwriting OR 1's allocations and creating overlapping appointments.

# Why Match Territory Does Not Prevent This:

Administrators often expect the Match Territory work rule to isolate territory optimization. While the rule behaves correctly, it governs policy scope rather than database transactions:

The resulting overlap is not a policy violation or misconfiguration. It is a data race condition where two independent processes schedule the same physical resources into identical time windows.