Docs-as-code help site

Use R pipelines without an Internet connection

DBP uses the Internet to download R and individual R packages. First, R downloads when DBP is installed on the ETL server. Next, packages download when the DBP Engine runs a pipeline. If your environment doesn't allow that, following are solutions.

Install R

A PowerShell script executes when DBP is installed that:

• Downloads and installs R
• Registers the R instance with DBP
• Downloads and installs the RODBC package

Script

DBPInstaller\SetupContent\DataBuild.Engine.InstallPackage\Install-R.ps1

Configuration file

Edit the rEngine scope of the install.config file to change the version, install path, library path, and RODBC version for your R instance.

Example

Use the table below to guide your RODBC package version selection:

If R is installed at D:\DataBuild\R\R-3.6.3, include a section like the following in the config file:

<scope name="rEngine">
    <variable name="Version" value="3.6.3" />
    <variable name="InstallPath" value="D:\DataBuild\R" />
    <variable name="LibraryPath" value="D:\DataBuild\R\Libraries" />
    <variable name="RODBCVersion" value="1.3-16" />
</scope>

No Internet connection

Register an R instance without an Internet connection

• Install R manually. Edit the install.config file to point to the install location (InstallPath).
• Install the RODBC package manually, preferably in a directory with the name of the version. Edit the install.config file LibraryPath and RODBCVersion. If you install the RODBC package per the configuration in the example above, the RODBC library is installed to D:\DataBuild\R\Libraries\RODBC\1.3-16\RODBC.
• Run the R installation script to configure the metadata for the R environment so Dataset Designer and the Engine know about the manually installed R instance.

You can also install the ODBC package instead of or in addition to the RODBC package. To use the ODBC package, enable the EnableOdbcPackageForR toggle in PlatformFeatureToggle.

Install packages

When an R pipeline is executed by the DBP Engine, the R script is modified with additional commands to download packages according to the metadata for the pipeline. In Dataset Designer, leave the package dependency list empty so the Engine doesn't download packages.

Following are two methods to manage your R packages without using built-in DBP support:

Set up a CRAN mirror

Set up a mirror of CRAN in your network (see CRAN documentation). Use the install.packages command in your R script to retrieve packages. Use the repos option to point to your mirror.

Install manually

Manually install any required packages into some location on the ETL server, such as D:\DataBuild\R\SiteLibrary. At the beginning of your R script, set the libPaths variable to point to that library location.

Allow access to sites

If DBP is allowed some Internet access, allow access to the following sites:

• https://cran.r-project.org
• https://cloud.r-project.org
• https://packagemanager.posit.co