Search queue saturation causes request aborts after timeout

When searches wait too long in the search queue to acquire a processing permit (exceeding time_to_abort duration), they are automatically aborted. This typically occurs when queue is saturated with up to 1000 pending searches and concurrent processing capacity (available_parallelism × 4 by default) is fully utilized.

Monitor search queue depth (number waiting) and searches currently processing. When queue consistently fills or searches are aborted, consider increasing search parallelism factor at startup or scaling horizontally. Track metrics: queue capacity (fixed at 1000), searches waiting, and searches currently being processed.