Problem

A cluster takes a long time to launch and displays an error message similar to the following:

Cluster is running but X nodes could not be acquired

Cause

Provisioning an Azure VM typically takes 2-4 minutes, but if all the VMs in a cluster cannot be provisioned at the same time, cluster creation can be delayed. This is due to Azure Databricks having to reissue VM creation requests over a period of time.

Solution

If a cluster launches without all of the nodes, Azure Databricks automatically tries to acquire the additional nodes and will update the cluster once available.

To workaround this, you should configure a cluster with a bigger instance type and a smaller number of nodes.