Upgrading Tecton SDK on Notebook Clusters

This guide includes instructions for updating existing Notebook clusters to a different Tecton SDK version. Supported SDK versions can be found here.

Upgrading Databricks Notebooks

In the Databricks UI, go to Clusters -> your current Tecton Notebook Cluster.

Replace Libraries and Jars in the new cluster

In the Cluster configuration page, go to the Libraries tab and uninstall the tecton and tecton-udfs jar.

Install tecton package from PyPi

Click Install New
Select PyPI under Library Source
Set Package to tecton and specify the new version. For example, if upgrading to version 0.2.7 specify tecton==0.2.7.

Install Tecton UDFs jar

Click Install New
Select DBFS/S3 under Library Source
Set File Path to s3://tecton.ai.public/pip-repository/itorgation/tecton/{tecton_version}/tecton-udfs-spark-3.jar and replace tecton_version with the new SDK version. For example, s3://tecton.ai.public/pip-repository/itorgation/tecton/0.2.7/tecton-udfs-spark-3.jar

Restart your Cluster

Click Restart
Connect a notebook to the cluster and run the following to confirm the new version.
```
import tecton
print(tecton.__version__)
```

Upgrading EMR Notebooks

The following instructions are for upgrading to a specific Tecton SDK version.

Best Practices

We highly recommend passing the Tecton SDK version in as an argument to the setup script. If a specific version is not pinned, the script will pull the latest stable version of Tecton. This may cause the Tecton SDK to be upgraded to the latest major or minor release unintentionally on a cluster restart.

In the EMR UI, go to Clusters -> your Tecton Notebook Cluster
Clone or Restart your cluster
Set the bootstrap script location to s3://tecton.ai.public/install_scripts/setup_emr_notebook_cluster_v2.sh, if it is not already pointing to that.
Change the argument passed in to the desired new version number, example 0.2.7.
Connect a notebook to the cluster and run the following to confirm the new version.
```
import tecton
print(tecton.__version__)
```