[ML] Add Telemetry for models without adaptive allocations #129161
Conversation
Added min and max allocations as attributes to the telemetry for trained models with adaptive allocations enabled. Added telemetry for models with adaptive allocations disabled or never set.
prwhelan
added
>enhancement
:ml
|
Hi @prwhelan, I've created a changelog YAML for you. |
|
Pinging @elastic/ml-core (Team:ML) |
|
|
||
| trainedModelsCurrentAllocations += trainedModelAssignment.totalCurrentAllocations(); | ||
| if (trainedModelAssignment.getAdaptiveAllocationsSettings() == null) { | ||
| trainedModelsFixedAllocations += trainedModelAssignment.totalCurrentAllocations(); |
There was a problem hiding this comment.
Here and in line 518 the code is summing the number of allocations from all deployments that do not use adaptive allocations. A single deployment could have 10 allocations and we wouldn't know if the user has 10 deployments with 1 allocation or 1 deployment with 10.
I think counting the number of deployments would be more meaningful
There was a problem hiding this comment.
Yeah that is a good point, we can just do an easy +1 to count the deployments
| "es.ml.trained_models.deployment.fixed_allocations.current", | ||
| "Sum of current trained model allocations that do not use adaptive allocations (either enabled or disabled)", | ||
| "allocations", | ||
| () -> new LongWithAttributes(trainedModelAllocationCounts.trainedModelsFixedAllocations, isMasterMap) |
There was a problem hiding this comment.
Can the project type be added to the attribute map? If there are different rules for different project types it would be useful to split the data that way
There was a problem hiding this comment.
I don't think so? It looks like it comes from serverless.project_type which isn't available here. We could move this metric to serverless, or we can use ES|QL magic to pull in the project type from other metrics via the project id.
It's possible this will get automatically added when running in serverless.
4 participants
Added min and max allocations as attributes to the telemetry for trained models with adaptive allocations enabled.
Added telemetry for models with adaptive allocations disabled or never set.
Verified on QA:
