User Telemetry

HPE Machine Learning Inferencing Software automatically collects and reports anonymous customer metrics. This data is used to improve the product and provide better support to users.

Telemetry Data #

The following information is periodically collected from the MLIS Controller.

Controller Details #

Category	Details
Category Identifiers	Unique, randomly generated ID for current database
Controller Info	Version of the controller
User Data	Number of user accounts
Registry Data	- Total number of registries - Number of registries of each type
Model Data	Number of packaged models
Deployment Data	- Number of deployments - Number of deployment tokens

Packaged Model Details #

Detail	Description
Registry Type	Type of registry for each packaged model
Model Format	Format of each packaged model
Required Resources	Resources required for each packaged model
Model Image/URI	Image or URI of each packaged model

Deployment Details #

Detail	Description
Autoscaling Requirements	Autoscaling requirements for each deployment
Model Image/URI	Image or URI for each deployment

Recorded Events #

Event Type	Details
Errors	Error messages reported on a deployment
Controller Issues	Unexpected controller shutdowns
Deployment Actions	- Creation - Updates - Pausing - Resuming - Deletion
Performance Metrics	Elapsed time between deploying a model and the deployment becoming ready for use

Disable Telemetry #

Telemetry can be disabled via helm upgrade at any time by specifying --set master.telemetry_enabled=false.

helm upgrade mlis \
   --set master.telemetry_enabled=false \
   --set 'global.imagePullSecrets[0].name=regcred' \
   --set 'global.imagePullSecrets[1].name=hpe-mlis-registry' \
   --set imageRegistry=hub.myenterpriselicense.hpe.com/hpe-mlis/<LOWERCASE_SKU> \
   --set defaultPassword=<CREATE_ADMIN_PASSWORD> \
   --values values.yaml \ # Required if you have custom configurations
   <SKU>_aioli-helm-chart_1.1.0.tgz

Article Summarization