From 4e52e2578345024521c6438f3ae4f7bc30c5b835 Mon Sep 17 00:00:00 2001
From: selmazrg <zargaselma@gmail.com>
Date: Mon, 19 Jun 2023 14:09:26 +0200
Subject: [PATCH] feat(sparkR): integration of sparkr

---
 roles/spark/client/tasks/install.yml             | 16 ++++++++++++++++
 .../common/templates/sparkr-shell-command.j2     |  7 +++++++
 tdp_vars_defaults/spark/spark.yml                |  3 +++
 tdp_vars_defaults/spark3/spark3.yml              |  3 +++
 4 files changed, 29 insertions(+)
 create mode 100644 roles/spark/common/templates/sparkr-shell-command.j2

diff --git a/roles/spark/client/tasks/install.yml b/roles/spark/client/tasks/install.yml
index 2a52d2dd..df0aced5 100644
--- a/roles/spark/client/tasks/install.yml
+++ b/roles/spark/client/tasks/install.yml
@@ -67,3 +67,19 @@
     owner: root
     group: root
     mode: "755"
+
+- name: "Render /usr/bin/{{ spark_version }}R-shell command"
+  ansible.builtin.template:
+    src: sparkr-shell-command.j2
+    dest: "/usr/bin/{{ spark_version }}r-shell"
+    owner: root
+    group: root
+    mode: "755"
+  when: (spark_version == "spark3") and spark_enable_r
+
+- name: Create symbolic link to R library
+  ansible.builtin.file:
+    src: "{{ spark_install_dir }}/R/lib/SparkR"
+    dest: "/usr/lib64/R/library/SparkR"
+    state: link
+  when: (spark_version == "spark3") and spark_enable_r
diff --git a/roles/spark/common/templates/sparkr-shell-command.j2 b/roles/spark/common/templates/sparkr-shell-command.j2
new file mode 100644
index 00000000..a35e0ca6
--- /dev/null
+++ b/roles/spark/common/templates/sparkr-shell-command.j2
@@ -0,0 +1,7 @@
+#!/usr/bin/env bash
+
+
+export YARN_CONF_DIR=/etc/hadoop/conf
+
+
+/opt/tdp/spark3/bin/sparkR "$@"
diff --git a/tdp_vars_defaults/spark/spark.yml b/tdp_vars_defaults/spark/spark.yml
index f93a225b..ca890b83 100644
--- a/tdp_vars_defaults/spark/spark.yml
+++ b/tdp_vars_defaults/spark/spark.yml
@@ -32,6 +32,9 @@ hbase_install_dir: "{{ spark_root_dir }}/hbase"
 # Spark pid directories
 spark_pid_dir: /var/run/spark
 
+# SparkR
+spark_enable_r: false
+
 #Spark logging configuration
 # Root logger should be: [RFA | DRFA]
 spark_root_logger: RFA
diff --git a/tdp_vars_defaults/spark3/spark3.yml b/tdp_vars_defaults/spark3/spark3.yml
index 9f70449c..5f7d076e 100644
--- a/tdp_vars_defaults/spark3/spark3.yml
+++ b/tdp_vars_defaults/spark3/spark3.yml
@@ -27,6 +27,9 @@ spark_hs_conf_dir: "{{ spark_conf_dir }}/conf.hs"
 # Spark pid directories
 spark_pid_dir: /var/run/spark3
 
+# SparkR
+spark_enable_r: false
+
 #Spark3 logging configuration
 # Root logger should be: [RFA | DRFA]
 spark_root_logger: RFA