Don't hard-code location of cognitive services #25
Previously we were assuming that the cognitive services called in the
As a tangential change, this commit also removes the ability to provide
The text was updated successfully, but these errors were encountered:
Previously we were assuming that the cognitive services called in the spark jobs are always deployed in WESTUS which leads to failed calls when the resources are set up somewhere else, e.g. in EASTUS. This change passes the resource group location in which the cognitive services were created through to spark so that we can format the appropriate base URL for the services. As a tangential change, this commit also removes the ability to provide cognitive services tokens at deploy time. We already have the ability to set up new cognitive services accounts during the fortis-deploy.sh script so we might as well leverage that to ensure that the cognitive services tokens and deployment locations don't risk getting out of sync which will lead to hard to debug errors for the user. This changes also has some related benefits, e.g. making billing more straight forward since all Fortis resources will be located in the same resource group by construction.