Potential Conflict Between `max-tasks-per-child` and Concurrent Task Processing?

[jiangbaihe]

Background: I'm exploring ways to prevent memory leaks in TaskIQ workers and noticed two relevant parameters:

1. max-async-tasks (default: 100) - Sets the maximum concurrent tasks per worker
2. max-tasks-per-child - Recycles workers after completing N tasks  Hypothetical Scenario: If I set max-tasks-per-child=1 (to aggressively recycle workers) while keeping the default max-async-tasks=100, I'm concerned this might create an edge case:
3. A worker starts processing multiple tasks concurrently (up to 100)
4. When any single task completes, the worker might meet the recycling condition
5. This could potentially terminate other in-progress tasks  Questions for Clarification:
6. Is this understanding correct, or does TaskIQ have safeguards against this?
7. How does the system count completed tasks when dealing with concurrent execution?

• Does it wait for all concurrent tasks to finish before counting?
• Or does completion of any single task trigger the count?

3. If this is a valid concern, would you recommend:

• Setting max-async-tasks=1 when using max-tasks-per-child?
• Or is there another preferred pattern for memory leak prevention?