none
RPackage library exception (error 1000) RRS feed

  • Question

  • Hi,

    I have the following code in an Execute R Module.

    ---------------------------------------------------------------------------

    # Input
    data1 <- maml.mapInputPort(1) # Qualitative with 8 variables

    install.packages("src/graphics.zip", lib.loc = ".", repos = NULL, verbose = TRUE)
    install.packages("src/grDevices.zip", lib.loc = ".", repos = NULL, verbose = TRUE)
    install.packages("src/stats.zip", lib.loc = ".", repos = NULL, verbose = TRUE)
    install.packages("src/utils.zip", lib.loc = ".", repos = NULL, verbose = TRUE)
    install.packages("src/MASS.zip", lib.loc = ".", repos = NULL, verbose = TRUE)
    success <- library("MASS", lib.loc = ".", logical.return = TRUE, verbose = TRUE)
    library(MASS)

    mca <- mca(data1, nf = 10)

    mca1 <- data.frame(mca$rs)

    # Output
    maml.mapOutputPort("mca1");

    -------------------------------------------------------------------------------

    When I execute I am getting the following error:

    RPackage library exception: Attempting to obtain R output before invoking execution process. (Error 1000)

    But it is working fine in RStudio.

    I also have a node that does the same process and it works without errors. I have executed it several times, sometimes it has worked for me and then it has returned error.

    Please let me know what the issue is.

    With regards,

    Celia

    Wednesday, December 20, 2017 10:30 AM

All replies

  • Hi,

    I think the packages you need are already pre-installed, so you do not need do the install.packages().

    https://msdn.microsoft.com/en-us/library/mt741980.aspx?f=255&MSPPError=-2147217396#bkmk_M

    I was able to get it to work in the notebook environment:

    library(MASS)

    Regards,
    Jaya



    Wednesday, December 20, 2017 6:06 PM
  • Thank you very much for the reply, Jaya

    It's true, the package is preinstalled. I have removed the installation lines of code from the package but the problem persists.

    I have tried a thousands things and I have executed another thousand, but I can not solve it.

    The strange thing is that the next node with the same code is still working, and the only difference between de two nodes is the number of rows of the input.

    The node that gives error has more than 80K rows, and the node that runs has more than 10K rows.

    Is there any type of row limit in the 'mca' function of the MASS package in ML Studio?

    Because in R-Studio it executes without problem with the 80K rows.

    With regards,

    Celia

    Friday, January 5, 2018 10:37 AM
  • Hi,

    Yes, there is a limit on the data set size of 10GB within AML Studio: https://docs.microsoft.com/en-us/azure/machine-learning/studio/faq

    For more flexibility on dataset size and VM requirements, it might be easier to run your code using Azure ML Workbench: https://docs.microsoft.com/en-us/azure/machine-learning/studio/faq

    Regards,
    Jaya 

    Friday, January 5, 2018 2:29 PM
  • Hi,

    I have optimized the code and now my experiment in ML Studio works without problem. But when running the web service in data factory I get error 1000 again and again.

    Why if my experiment works the automation returns error?

    Regards,

    Celia

    Monday, January 8, 2018 1:10 PM
  • Hi,

    Did you check your web service from within Azure ML Studio to see if it works?

    https://docs.microsoft.com/en-us/azure/machine-learning/studio/walkthrough-5-publish-web-service

    Regards,
    Jaya


    Monday, January 8, 2018 2:56 PM
  • Hi,

    Ok.

    I obtain this: Unable to finish 'experiment' test. The model had exceeded the memory quota assigned to it.

    Is it possible to increase this memory limit? (2560 MB)

    Regards,

    Celia

    Monday, January 8, 2018 3:25 PM
  • Hi,

    Unfortunately, there is no customization of the memory size that can be done. This is good feedback for the product team: https://feedback.azure.com/forums/257792-machine-learning

    The new Azure ML workbench however gives the user control over the context and is customization, you could try and use that instead: https://docs.microsoft.com/en-us/azure/machine-learning/preview/quickstart-installation

    Regards,
    Jaya

    Monday, January 8, 2018 5:14 PM
  • Hi,

    My experiment is a clustering algorithm that uses a lot of data (the input is more than 500K rows and 100 columns), although at the time of doing the calculations both the rows and the columns used are less. I have reduced the number of nodes as much as possible. What else could I do to optimize it?

    The strange thing is that this experiment has been running for months and had not given any problem. Why did it give problems two weeks ago if the data number has not changed?

    On the other hand, regarding Azure ML Workbench, is it possible to export an experiment from Azure ML Studio to Azure ML Workbench? and is it possible to connect to a SQL database?

    Sorry for the inconvenience.

    Regards,

    Celia

    Tuesday, January 9, 2018 9:36 AM
  • Hi,

    Perhaps someone from the engineering team can respond on whether something changed within the Azure ML Studio environment.

    If your code is in Python, you can use the same script within Azure ML Workbench. It does look like you can connect to a Database as shown below from within Azure ML Workbench.

    Regards,
    Jaya

    Tuesday, January 9, 2018 2:43 PM
  • Hello, I had the same problem. In my case, switching R version from "Microsoft R Open" to "CRAN" (in the context menu related to "Execute R Script" module) helped.

    Kind regards,

    JD

    Saturday, September 28, 2019 2:01 PM