…on Count (elastic#143077)

The trained model stats API was computing `required_native_memory_bytes` incorrectly. Since elastic#98139, memory estimation has depended on the number of allocations, but the code used the total allocation count across all deployments instead of each deployment’s own count. As a result, when multiple NLP models were deployed with different allocation counts, every model’s `required_native_memory_bytes` was based on the same summed value, so changing allocations for one model incorrectly changed the reported memory for others. This only affected the Stats API output, not actual deployment behavior.

The fix computes `required_native_memory_bytes` per deployment using each deployment’s allocation count. `TransportGetTrainedModelsStatsAction` now passes `Map<String, AssignmentStats>` into `modelSizeStats()` instead of a summed allocation count, and `modelSizeStats()` emits per-deployment entries keyed by `deploymentId` with the correct `numberOfAllocations`. `GetTrainedModelsStatsAction.Response.Builder.build()` looks up model size stats by `deploymentId` first and falls back to `modelId` for undeployed or non-PyTorch models. Unit tests were added to cover per-deployment resolution, undeployed models, and the fallback path.


Fixes elastic#107831

…on Count (elastic#143077)

The trained model stats API was computing `required_native_memory_bytes` incorrectly. Since elastic#98139, memory estimation has depended on the number of allocations, but the code used the total allocation count across all deployments instead of each deployment’s own count. As a result, when multiple NLP models were deployed with different allocation counts, every model’s `required_native_memory_bytes` was based on the same summed value, so changing allocations for one model incorrectly changed the reported memory for others. This only affected the Stats API output, not actual deployment behavior.

The fix computes `required_native_memory_bytes` per deployment using each deployment’s allocation count. `TransportGetTrainedModelsStatsAction` now passes `Map<String, AssignmentStats>` into `modelSizeStats()` instead of a summed allocation count, and `modelSizeStats()` emits per-deployment entries keyed by `deploymentId` with the correct `numberOfAllocations`. `GetTrainedModelsStatsAction.Response.Builder.build()` looks up model size stats by `deploymentId` first and falls back to `modelId` for undeployed or non-PyTorch models. Unit tests were added to cover per-deployment resolution, undeployed models, and the fallback path.


Fixes elastic#107831

…cations

* upstream/main:
  Warn on API key version mismatch (elastic#143127)
  Fixed wrong malformed value ordering in synthetic source tests (elastic#143187)
  [ML] Fix: required_native_memory_bytes Calculated with Wrong Allocation Count (elastic#143077)
  Add configureBenchmarkLogging calls across the various benchmarks (elastic#143185)
  Mute org.elasticsearch.xpack.esql.CsvIT test {csv-spec:k8s-timeseries-avg-over-time.Avg_over_time_aggregate_metric_double_implicit_casting} elastic#143292
  Give system role permission to invoke shard refresh (elastic#143190)
  Mute testSyntheticSourceWithTranslogSnapshot (elastic#143260)
  Adds ResumeInfo Tests (elastic#142769)
  Use a static method to configure benchmark logging (elastic#143056)
  add connectors release notes (elastic#142884)
  Add CI triage guidance for AI agents (elastic#142994)
  ESQL: Data sources: ZSTD, BZIP2 (elastic#143228)
  [ES|QL] Channels issue when an agg is called with the same field (elastic#142180) (elastic#142269)
  Add support for project routing in reindex requests (elastic#142240)

…on Count (#143077) (#143295)

The trained model stats API was computing `required_native_memory_bytes` incorrectly. Since #98139, memory estimation has depended on the number of allocations, but the code used the total allocation count across all deployments instead of each deployment’s own count. As a result, when multiple NLP models were deployed with different allocation counts, every model’s `required_native_memory_bytes` was based on the same summed value, so changing allocations for one model incorrectly changed the reported memory for others. This only affected the Stats API output, not actual deployment behavior.

The fix computes `required_native_memory_bytes` per deployment using each deployment’s allocation count. `TransportGetTrainedModelsStatsAction` now passes `Map<String, AssignmentStats>` into `modelSizeStats()` instead of a summed allocation count, and `modelSizeStats()` emits per-deployment entries keyed by `deploymentId` with the correct `numberOfAllocations`. `GetTrainedModelsStatsAction.Response.Builder.build()` looks up model size stats by `deploymentId` first and falls back to `modelId` for undeployed or non-PyTorch models. Unit tests were added to cover per-deployment resolution, undeployed models, and the fallback path.


Fixes #107831

…on Count (#143077) (#143294)

The trained model stats API was computing `required_native_memory_bytes` incorrectly. Since #98139, memory estimation has depended on the number of allocations, but the code used the total allocation count across all deployments instead of each deployment’s own count. As a result, when multiple NLP models were deployed with different allocation counts, every model’s `required_native_memory_bytes` was based on the same summed value, so changing allocations for one model incorrectly changed the reported memory for others. This only affected the Stats API output, not actual deployment behavior.

The fix computes `required_native_memory_bytes` per deployment using each deployment’s allocation count. `TransportGetTrainedModelsStatsAction` now passes `Map<String, AssignmentStats>` into `modelSizeStats()` instead of a summed allocation count, and `modelSizeStats()` emits per-deployment entries keyed by `deploymentId` with the correct `numberOfAllocations`. `GetTrainedModelsStatsAction.Response.Builder.build()` looks up model size stats by `deploymentId` first and falls back to `modelId` for undeployed or non-PyTorch models. Unit tests were added to cover per-deployment resolution, undeployed models, and the fallback path.


Fixes #107831

…on Count (elastic#143077)

The trained model stats API was computing `required_native_memory_bytes` incorrectly. Since elastic#98139, memory estimation has depended on the number of allocations, but the code used the total allocation count across all deployments instead of each deployment’s own count. As a result, when multiple NLP models were deployed with different allocation counts, every model’s `required_native_memory_bytes` was based on the same summed value, so changing allocations for one model incorrectly changed the reported memory for others. This only affected the Stats API output, not actual deployment behavior.

The fix computes `required_native_memory_bytes` per deployment using each deployment’s allocation count. `TransportGetTrainedModelsStatsAction` now passes `Map<String, AssignmentStats>` into `modelSizeStats()` instead of a summed allocation count, and `modelSizeStats()` emits per-deployment entries keyed by `deploymentId` with the correct `numberOfAllocations`. `GetTrainedModelsStatsAction.Response.Builder.build()` looks up model size stats by `deploymentId` first and falls back to `modelId` for undeployed or non-PyTorch models. Unit tests were added to cover per-deployment resolution, undeployed models, and the fallback path.


Fixes elastic#107831