USER MANUALS

Deployment

The cluster.sh script, in the root folder of the distribution, automates the deployment of the Denodo Embedded MPP on Kubernetes using a Helm chart.

Important

To run the Embedded MPP on Windows:

  • You need a Bash compatible shell such as Cygwin or Git Bash installed or use Windows Subsystem for Linux (WSL).

  • Check if the environment variable HADOOP_HOME is set on this computer, since Hadoop is required by cluster.sh to transparently manage the encryption of all user-provided credentials.

    If HADOOP_HOME is not set:

    1. Create a directory, e.g., <DENODO_HOME>hadoop_win_utils.

    2. Create a directory named bin inside the new directory, e.g., <DENODO_HOME>hadoop_win_utilsbin.

    3. Set the environment variable HADOOP_HOME to point to <DENODO_HOME>\hadoop_win_utils.

    4. Copy the content of the <DENODO_HOME>\dll\vdp\winutils directory to %HADOOP_HOME%\bin.

The cluster.sh script has these prerequisites:

  1. kubectl, Helm v3 and Java are required.

  2. Configure the authentication to the Container image registry where the Denodo Embedded MPP container images reside.

    See the Container Image Registry Credentials section.

Let’s see now how to deploy the Denodo Embedded MPP:

The Denodo Embedded MPP is distributed as a Helm chart. However, we still provide the cluster.sh script on top of the Helm chart that transparently manages the encryption of all credentials supplied by the user. To do this, the Hadoop credential provider framework is used to create a keystore file, creds.jceks, to avoid using clear values (e.g. the S3 Access Key Id and the S3 Secret Access Key).

cluster.sh deploy [OPTIONS]

COMMON OPTIONS

  • --presto-password: This option sets the password for the presto user, instead of using the default: pr3st%.

    If this password is not specified in the command line, cluster.sh deploy will prompt for it, keeping passwords out of the bash history. Otherwise, the presto.coordinator.passwordAuth.prestoPassword supplied in the values.yaml file will be used.

  • --credstore-password: Password to protect the encrypted credentials file, creds.jceks, which will store all the credentials supplied by the user.

    If this password is not specified in the command line, cluster.sh deploy will prompt for it, keeping passwords out of the bash history.

  • --metastore-password: Password for the database of the Embedded Hive Metastore.

    If this password is not specified in the command line, cluster.sh deploy will prompt for it, keeping passwords out of the bash history. Otherwise, the metastore.connectionPassword supplied in the values.yaml file will be used.

  • --register: With this option the deployment process includes a final step that creates a special data source in Denodo called ‘embedded_mpp’. It also configures the Denodo query optimizer to use the Denodo Embedded MPP to accelerate queries.

    This feature requires the Denodo subscription bundle Enterprise Plus.

The rest of the options of the cluster.sh script depends on the Object Storage you want to access with the Denodo Embedded MPP. See next section.

Add feedback