The core jstest jstests/core/timeseries/query/agg_stage_coverage.js fails intermittently in balancer-enabled sharded passthroughs with:

assert failed : $planCacheStats expected to return documents on timeseries collections.
  @jstests/core/timeseries/query/agg_stage_coverage.js:349:11
  @jstests/core/timeseries/query/agg_stage_coverage.js:335:36

Root cause. The test seeds the plan cache with a query that has two competing plans, then later iterates over all aggregation stages and asserts that [{$planCacheStats: {]}} returns at least one entry.

In suites running with the balancer enabled, the balancer can perform a moveCollection (implemented as resharding) on test.agg_stage_coverage at any time. Migrating/moving a collection clears the plan cache on the owning shard, so by the time the loop reaches the $planCacheStats case the cache is empty and the assert(result.length > 0) fails.

Observed sequence from the failing run:

• 03:50:43 – collection created, documents inserted, plan cache seeded.
• 03:50:44.4 – 45.66 – balancer runs moveCollection ("provenance":"balancerMoveCollection",
  "Resharding complete" at 03:50:45.661), clearing the plan cache.
• 03:50:45.686 – the loop reaches $planCacheStats, sees an empty cache, and the assertion fails.

The IngressRequestRateLimitExceeded messages in the log are the balancer's own throttled internal config reads and are unrelated to the failure.