0 votes
Hi,

I'm working on a machine with no connectivity to the internet and integrated R using this link: https://doc.dataiku.com/dss/latest/installation/r.html#case-2-if-your-dss-server-does-not-have-internet-access

For the CRAN mirror, I have given the location of the local directory where packages are placed:

file:///opt/dss/R-packages/R-packages/

However, it isn't able to pick up the packages from the local directory and is throwing this error:

Updating code environment according to spec ...
Installing from R packages list
Installing from "file:///opt/dss/dataiku-dss-5.1.0/dku-jupyter/R", "file:///opt/dss/R-packages/R-packages/"
Installing into /var/opt/dss/code-envs/R/R-test/R.lib
[1] "known library paths:"
[1] "/var/opt/dss/code-envs/R/R-test/R.lib"
[2] "/opt/dss/dataiku-dss-5.1.0/R"         
[3] "/usr/lib64/R/library"                 
[4] "/usr/share/R/library"                 
[1] "already has : "
         dataiku    dataiku.spark   dataiku.spark2 dataiku.sparklyr
         "5.1.0"          "5.1.0"          "5.1.0"          "5.1.0"
            base             boot            class          cluster
         "3.5.1"         "1.3-20"         "7.3-14"        "2.0.7-1"
       codetools         compiler         datasets          foreign
        "0.2-15"          "3.5.1"          "3.5.1"         "0.8-70"
        graphics        grDevices             grid       KernSmooth
         "3.5.1"          "3.5.1"          "3.5.1"        "2.23-15"
         lattice             MASS           Matrix          methods
       "0.20-35"         "7.3-50"         "1.2-14"          "3.5.1"
            mgcv             nlme             nnet         parallel
        "1.8-24"        "3.1-137"         "7.3-12"          "3.5.1"
           rpart          spatial          splines            stats
        "4.1-13"         "7.3-11"          "3.5.1"          "3.5.1"
          stats4         survival            tcltk            tools
Checking installed packages ...
         "3.5.1"         "2.42-3"          "3.5.1"          "3.5.1"
           utils
         "3.5.1"
Package not installed: dplyr
Package not installed: httr
Package not installed: RJSONIO
Package not installed: gtools
Package not installed: base64enc
Package not installed: curl
Package not installed: IRkernel
Installing packages: dplyr httr RJSONIO gtools base64enc curl IRkernel
Error in read.dcf(file = tmpf) : cannot open the connection
Calls: install.packages -> available.packages -> read.dcf
In addition: Warning message:
In read.dcf(file = tmpf) :
  cannot open compressed file '/opt/dss/R-packages/R-packages/src/contrib/PACKAGES', probable reason 'No such file or directory'
Execution halted
OK

 

 

The packages needed by R are placed in the directory: /opt/dss/R-packages/R-packages/

 

Could you please advise on how to move forward with this?

 

Thanks
by
Roth,

I have successfully worked with miniCRAN, a packaged created and used by Microsoft's Revolution Analytics to mirror packages necessary to install R in a disconnected environment.

https://cran.r-project.org/web/packages/miniCRAN/

First, install miniCRAN in your Internet-connected environment and follow the vignettes on how to mirror the set of packages. miniCRAN will then layout the packages in a CRAN-compatible format. Once complete, move the folder from this environment into your disconnected Dataiku environment. After the folder is moved, you will need to point Dataiku at the location of this new folder for the redirection of all CRAN calls. This is located in the Administration --> Settings --> Misc section. There you will find the CRAN mirror URL. You will then place the local mapped folder location in the CRAN mirror URL location. Dataiku expects the following format for local folders:
file://<parent folder>/<sub folder>/miniCRAN

Note: the Dataiku user on the server must have access to this folder in order for it to work correctly. Hopefully this helps.

Grant

1 Answer

0 votes
At the moment, the download-R-packages script can only be used for the base installation of DSS, not for code envs.

For code envs, you'll need to create a real CRAN mirror. There are various R packages that can help you for that.
by
1,249 questions
1,277 answers
1,450 comments
11,801 users

©Dataiku 2012-2018 - Privacy Policy