Search

Developer Workflow Guide

Developer Workflow Guide

10.4.1
- 10.5.6
- 10.5.3
- 10.5.2
- 10.5
- 10.4.0
- 10.2.2
- 10.2.1

Back Next

Advanced Options

Advanced Options

Configure advanced options such as automatic termination and on-demand instances.

The following table describes advanced options that you can set for an AWS Databricks cluster:

Property	Description
Enable autoscaling local storage	Enables Databricks to monitor available disk space on worker nodes and automatically add additional EBS volumes.
EBS Volume Type	The type of volume that Databricks can add to cluster nodes. Set this property when you enable autoscaling.
Number of Volumes	The number of volumes to provision for each instance. Enter a value between 0 and 10. Set this property when you enable autoscaling and configure EBS volume type.
Size in GB	The size in gigabytes of each EBS volume. Set this property when you enable autoscaling and configure EBS volume type.
Auto Termination	Enables automatic termination of the cluster.
Auto Terminate Time	Terminates the cluster after it is inactive for the specified number of minutes. Enter a value between 10 and 10,000. If you do not configure this, or if you set to 0, the cluster will not automatically terminate.
On-Demand/Spot Composition	The number of on-demand nodes. Enter a value between 0 and the number of worker nodes set in General Options. Any remaining worker nodes will be spot instances. On-demand nodes are always available to use. Spot instances might terminate running jobs if they become unavailable. The driver node is always an on-demand node. Set this property when you enable Spot fall back to On-Demand . Default is 1.
Spot fall back to On-Demand	Enables on-demand instances to be used as a fallback. If you are using spot instances and the market price for spot instances surges above your spot bid price, AWS terminates the spot instances. When you enable this property, on-demand instances are used in place of the spot instances when they terminate.
Availability Zone	The AWS cluster availability zone. Default is us-east-1e.
Spot Bid Price	The maximum percent of the on-demand instance price that you bid on spot instances. Spot instances are priced as a percentage of the on-demand price and are not always available. If the market price for spot instances surges above the bid price set here and you do not enable Spot fall back to On-Demand , AWS terminates the spot instance. Default is 100%.
IAM Role ARN	The instance profile ARN (Amazon Resource Name) that corresponds to the AWS IAM (Identity and Access Management) role. Copy the value from the AWS console in the following format: arn:aws:iam::<account-id>:instance-profile/<role-name> IAM roles allow you to access data from Databricks clusters. Add new IAM roles in the Administrator tool.
Spark Configurations	Performance configurations for the Databricks Spark engine. Enter key-value pairs in the following format: key1='value1' key2='value2'. You can also provide a path to a file that contains the key-value pairs.
Environment Variables	Environment variables that you can configure for the Databricks Spark engine. Enter key-value pairs in the following format: key1='value1' key2='value2'
Cluster Tags	Labels that you can assign to resources for tracking purposes. Enter key-value pairs in the following format: <key1>=<value1>,<key2>=<value2>. You can also provide a path to a local file that contains the key-value pairs. Use the following format: file:\\<file path>
SSH Public Key	The SSH public key to log into the driver and worker instances if you enable SSH. Copy the value from the Databricks console.
Cluster Log Conf	The location to deliver logs for long-term storage. If configured, the Databricks Spark engine will deliver the logs every five minutes. Provide the path to DBFS.
Init Scripts	The location where you store init scripts. You can enter multiple destinations. The scripts are run sequentially in the order that you configure them. If you need to install additional Python libraries, specify the init script file location in this property. Use the following format: dbfs:/<path to init script>,dbfs:/<path to init script>

Advanced Properties for AWS Databricks

Watch

Comments

0 COMMENTS

We’d like to hear from you! Log in to comment.