When we execute the R bnames demo, the sweep script runs an R script that installs the plyr and ggplot2 packages. However these packages are not getting installed on the headnode, because it doesn't run any parametric sweep tasks (when you create an HPC VM in Azure, the head node itself is offline, so it doesn't run any tasks).
Therefore, when we run the post processing script that requires plyr and ggplot2, the R script fails.
I think this did not show up during the demo creation because as I see in the screenshots, the head node was online, and therefore probably participated in the sweep job and installed the missing packages.
To resolve this issue we need to add the following line at the top of the post processing R script:
install.packages(c("plyr", "ggplot2"), repos=c("http://cran.wustl.edu", "http://cran.us.r-project.org"))
(under Day 2\3. Windows HPC Server 2012 Cluster\4. hpc-R-sweep-optional)
JohnWintellect
commented
9 years ago
When we execute the R bnames demo, the sweep script runs an R script that installs the plyr and ggplot2 packages. However these packages are not getting installed on the headnode, because it doesn't run any parametric sweep tasks (when you create an HPC VM in Azure, the head node itself is offline, so it doesn't run any tasks). Therefore, when we run the post processing script that requires plyr and ggplot2, the R script fails. I think this did not show up during the demo creation because as I see in the screenshots, the head node was online, and therefore probably participated in the sweep job and installed the missing packages.
To resolve this issue we need to add the following line at the top of the post processing R script: install.packages(c("plyr", "ggplot2"), repos=c("http://cran.wustl.edu", "http://cran.us.r-project.org"))
(under Day 2\3. Windows HPC Server 2012 Cluster\4. hpc-R-sweep-optional)