Hydrospheredata / hydro-serving

MLOps Platform
http://docs.hydrosphere.io
Apache License 2.0
271 stars 42 forks source link

how do you increase the number of instances of a Model? #232

Closed ravimadhusudhan8 closed 5 years ago

tidylobster commented 5 years ago

Hi, @ravimadhusudhan8. This depends on how you've instantiated serving. If you've done this with docker-compose, then this option is not available for you. If you've done this on the Kubernetes cluster, then you can scale a deployment. Currently application deployment have the following name: r{runtime_id}m{model_id}e{environment_id}. So, for example it might be something like r1m1e0. If you have multiple applications then this might be a little problematic since it's hard to identify which of the application has which name. We are aware of this and will improve the naming soon. Possible workaround for now is to identify which of the deployment you want to scale by looking at its pod logs and then apply the scaling:

$ kubectl scale --replicas=2 deployment/r1m1e0
ravimadhusudhan8 commented 5 years ago

Thanks for the feedback. Will work on using Kubernetes cluster and will apply your suggestions.