Task Distribution Challenges and Solutions in a Small Court System

The article abstracts tasks as a universal concept (e.g., delivery orders, dispute reviews) and presents a simplified task system diagram focusing on core components.

Key challenges in task distribution include:

Backlog caused by task inflow exceeding consumption speed.

Abnormal consumption where tasks are completed in ways that violate business rules.

Over‑consumption under high concurrency leading to resource waste.

Task Backlog : In the original allocation model, a timed job runs hourly, fetching all pending tasks, randomly selecting users, performing rule checks, and assigning locks. Processing a single task takes 1–2 seconds; with ~3000 daily tasks, the job can exceed its time window, resulting in missed executions and massive backlog.

Active Claim Model : The system was refactored to let qualified users actively claim tasks. Qualification checks are moved to the user module, decoupling from the distribution flow, reducing intrusive system messages.

Abnormal Consumption : After the claim model launch, some reviewers sold “votes”. To mitigate this, a user lock queue limits each user to a configurable number of visible tasks within a time window, preventing unlimited task switching.

Over‑Consumption : High traffic and eager reviewers cause tasks to be consumed faster than needed, wasting voting resources. The article models voting capacity (y), remaining votes (r), and conflict condition (r < y). Reducing user fatigue (F) can lower concurrent vote volume (Na), but may hurt user experience.

Task Lock Queue Logic (used in both models):

Fetch n tasks from the database into memory.

If tasks remain, proceed; otherwise stop.

Validate against the task lock queue.

If queue size > m, retry step 2; if size < m, add the user to the queue with a TTL.

Return the selected task.

If no task is available, return none.

Results show that after switching to the claim model and introducing user/task lock queues, task completion rate reached 100%, average processing time dropped from 20 hours to under 2 hours, and resource waste from concurrent conflicts was significantly reduced.

The article concludes with a recommendation to use factory‑plus‑strategy patterns for extensible task inflow modules, ensuring rapid and stable integration of new scenarios.