Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/jmlondon/nmfs_rsconnect

Dockerfile to mirror the NMFS RStudio::Connect server
https://github.com/jmlondon/nmfs_rsconnect

Last synced: 5 days ago
JSON representation

Dockerfile to mirror the NMFS RStudio::Connect server

Awesome Lists containing this project

README 

        
# nmfs_rsconnect

Dockerfile to test NMFS RStudio::Connect server for dependecies



## summary

The NMFS RStudio::Connect server is running on AWS with CentOS 7 as the operating system. CentOS 7 does not provide access to

more current versions of system libraries that are required for some R packages. This is especially true for the spatial

stack of system dependencies (e.g. PROJ, GEOS, GDAL) and R packages (e.g. sf, stars, terra, sp, raster). In addition to the

spatial stack, the gcc version provided by CentOS 7 is 4.8 and some R packages require newer versions of gcc to compile. Here

we use a Dockerfile to document the necessary setup. This Dockerfile can also be used to create a local Docker image for testing

installation of R packages prior to deployment on the NMFS RStudio::Connect server.



## setup

Here we start by pulling the base CentOS 7 image and setting up a few details for the build of R.



```dockerfile

FROM centos:7



LABEL maintainer="Josh M London "



ARG R_VERSION

ARG BUILD_DATE

ARG CRAN

ENV BUILD_DATE ${BUILD_DATE:-2020-09-11}

ENV R_VERSION=${R_VERSION:-4.1.1} \

    CRAN=${CRAN:-https://cran.rstudio.com} \ 

    LC_ALL=en_US.UTF-8 \

    LANG=en_US.UTF-8 

```



Several of the dependencies and many R packages rely on a current version of sqlite. We will need to build sqlite

from source. The following environment variable must be set for successful sqlite build



```dockerfile

ENV PKG_CONFIG_PATH=/usr/local/lib/pkgconfig

```



## development tools

The base OS does not come with many of the needed compiling and development tools we will rely on for compiling

various libraries and packages from source. We start by installing the "Development Tools" group.



```dockerfile

RUN yum -y groupinstall "Development Tools"

```



This installation, however, installs an outdated version of gcc. We could install a new version of gcc from

source, but there is an easier and less intrusive method. The approach described here is adapted from an

[article on the RStudio Connect Support site](https://support.rstudio.com/hc/en-us/articles/360006142673-RStudio-Connect-with-devtoolset-enabled).



We are going to rely on the newer `devtooset-10`. Note, this installs gcc version 10 along side the already installed gcc 4.8.



```dockerfile

RUN yum install -y centos-release-scl

RUN yum install -y devtoolset-10

```



## setup for RStudio::Connect

In many use cases, one could call `source scl_source enable devtoolset-10` from a command line to enable `devtoolset-10` and

have access to updated gcc version. However, in the case of RStudio::Connect, we need to ensure all compilation processes

during deployment/install of an application use `devtoolset-10`. RStudio::Connect enables this through the use of a 

Program Supervisor.



First, we'll want to create a directory to hold the scripts on the server



```bash

mkdir -p /opt/scripts

```



And, then, create a script file `devtools-enable.sh` with the following code:



```bash

#!/bin/bash



echo arguments: "$@" >&2

echo >&2



source scl_source enable devtoolset-10



exec "$@"

```



We meed to allow access to the file with



```bash

chmod 755 /opt/scripts/devtool-enable.sh

```



Finally, we need to edit the configuration file `/etc/rstudio-connect/rstudio-connect.gcfg` by 

adding the following entry



```bash

[Applications]

Supervisor = /opt/scripts/devtool-enable.sh

```



## setup for Docker

Since we're not actually running RStudio::Connect as part of Docker, we have do implement a slighty 

different approach or local Docker testing. We handle this by COPYing our `devtool-enable.sh` file

into our image and adding an ENTRYPOINT



```dockerfile

COPY devtools-enable.sh /usr/bin/devtools-enable.sh

RUN chmod +x /usr/bin/devtools-enable.sh

ENTRYPOINT [ "/usr/bin/devtools-enable.sh" ]

```



## additional dependencies

The following dependencies are libraries needed for successful compilation of the spatial dependencies

from source in the next section



```dockerfile

RUN yum install -y openssl-devel

RUN yum install -y epel-release

RUN yum install -y udunits2-devel

RUN yum install -y cpp 

RUN yum install -y libtiff-devel 

RUN yum install -y cmake3

RUN yum install -y libcurl-devel

```



## sqlite 3

The version of sqlite availble from the typical CentOS7 repositories is not compatible with our

spatial stack so we need to install the latest version from source.



```dockerfile

RUN curl -L -O https://www.sqlite.org/2021/sqlite-autoconf-3360000.tar.gz \

  && tar -zxvf sqlite-autoconf-3360000.tar.gz \

  && cd sqlite-autoconf-3360000 \

  && ./configure \

  && make \

  && make install

```



## proj

The PROJ library for spatial projection support



```dockerfile

RUN curl -L -O https://download.osgeo.org/proj/proj-7.2.1.tar.gz \

  && tar -zxvf proj-7.2.1.tar.gz \

  && cd proj-7.2.1 \

  && ./configure \

  && make \

  && make install \

  && projsync --all

```



## geos

The GEOS library for spatial functions and computation



```dockerfile

RUN curl -L -O http://download.osgeo.org/geos/geos-3.9.1.tar.bz2 \

  && tar -xvf geos-3.9.1.tar.bz2 \

  && cd geos-3.9.1 \

  && ./configure \

  && make \

  && make install

```



## gdal

The GDAL library for read/write, transformations, of spatial data (this will take relatively longer to build)



```dockerfile

RUN curl -L -O https://github.com/OSGeo/gdal/releases/download/v3.3.2/gdal-3.3.2.tar.gz \

  && tar -zxvf gdal-3.3.2.tar.gz \

  && cd gdal-3.3.2 \

  && ./configure --with-proj=/usr/local --with-sqlite3=/usr/local \

  && make -j 4\

  && make install

```



## R 4.1.1

The final step for our Docker image is to install the latest version of R so we can test package installation



```dockerfile

RUN curl -O https://cdn.rstudio.com/r/centos-7/pkgs/R-${R_VERSION}-1-1.x86_64.rpm \

  && yum -y install R-${R_VERSION}-1-1.x86_64.rpm \

  && ln -s /opt/R/${R_VERSION}/bin/R /usr/local/bin/R \

  && ln -s /opt/R/${R_VERSION}/bin/Rscript /usr/local/bin/Rscript \

  && echo "options(repos = c(CRAN = 'http://cran.rstudio.com'))" >> /opt/R/${R_VERSION}/lib/R/etc/Rprofile.site \

  && echo "TZ=Etc/UTC" >> /opt/R/${R_VERSION}/lib/R/etc/Renviron.site

```