diff --git a/dev/create-release/do-release-docker.sh b/dev/create-release/do-release-docker.sh
new file mode 100755
index 0000000000000..c1a122ebfb12e
--- /dev/null
+++ b/dev/create-release/do-release-docker.sh
@@ -0,0 +1,146 @@
+#!/usr/bin/env bash
+
+#
+# Licensed to the Apache Software Foundation (ASF) under one or more
+# contributor license agreements.  See the NOTICE file distributed with
+# this work for additional information regarding copyright ownership.
+# The ASF licenses this file to You under the Apache License, Version 2.0
+# (the "License"); you may not use this file except in compliance with
+# the License.  You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+#
+# Creates a Spark release candidate. The script will update versions, tag the branch,
+# build Spark binary packages and documentation, and upload maven artifacts to a staging
+# repository. There is also a dry run mode where only local builds are performed, and
+# nothing is uploaded to the ASF repos.
+#
+# Run with "-h" for options.
+#
+
+set -e
+SELF=$(cd $(dirname $0) && pwd)
+. "$SELF/release-util.sh"
+
+function usage {
+  local NAME=$(basename $0)
+  cat <<EOF
+Usage: $NAME [options]
+
+This script runs the release scripts inside a docker image. The image is hardcoded to be called
+"spark-rm" and will be re-generated (as needed) on every invocation of this script.
+
+Options are:
+
+  -d [path]   : required: working directory (output will be written to an "output" directory in
+                the working directory).
+  -n          : dry run mode. Performs checks and local builds, but do not upload anything.
+  -t [tag]    : tag for the spark-rm docker image to use for building (default: "latest").
+  -j [path]   : path to local JDK installation to use for building. By default the script will
+                use openjdk8 installed in the docker image.
+  -s [step]   : runs a single step of the process; valid steps are: tag, build, docs, publish
+EOF
+}
+
+WORKDIR=
+IMGTAG=latest
+JAVA=
+RELEASE_STEP=
+while getopts "d:hj:ns:t:" opt; do
+  case $opt in
+    d) WORKDIR="$OPTARG" ;;
+    n) DRY_RUN=1 ;;
+    t) IMGTAG="$OPTARG" ;;
+    j) JAVA="$OPTARG" ;;
+    s) RELEASE_STEP="$OPTARG" ;;
+    h) usage ;;
+    ?) error "Invalid option. Run with -h for help." ;;
+  esac
+done
+
+if [ -z "$WORKDIR" ] || [ ! -d "$WORKDIR" ]; then
+  error "Work directory (-d) must be defined and exist. Run with -h for help."
+fi
+
+if [ -d "$WORKDIR/output" ]; then
+  read -p "Output directory already exists. Overwrite and continue? [y/n] " ANSWER
+  if [ "$ANSWER" != "y" ]; then
+    error "Exiting."
+  fi
+fi
+
+cd "$WORKDIR"
+rm -rf "$WORKDIR/output"
+mkdir "$WORKDIR/output"
+
+get_release_info
+
+# Place all RM scripts and necessary data in a local directory that must be defined in the command
+# line. This directory is mounted into the image.
+for f in "$SELF"/*; do
+  if [ -f "$f" ]; then
+    cp "$f" "$WORKDIR"
+  fi
+done
+
+GPG_KEY_FILE="$WORKDIR/gpg.key"
+fcreate_secure "$GPG_KEY_FILE"
+$GPG --export-secret-key --armor "$GPG_KEY" > "$GPG_KEY_FILE"
+
+run_silent "Building spark-rm image with tag $IMGTAG..." "docker-build.log" \
+  docker build -t "spark-rm:$IMGTAG" --build-arg UID=$UID "$SELF/spark-rm"
+
+# Write the release information to a file with environment variables to be used when running the
+# image.
+ENVFILE="$WORKDIR/env.list"
+fcreate_secure "$ENVFILE"
+
+function cleanup {
+  rm -f "$ENVFILE"
+  rm -f "$GPG_KEY_FILE"
+}
+
+trap cleanup EXIT
+
+cat > $ENVFILE <<EOF
+DRY_RUN=$DRY_RUN
+SKIP_TAG=$SKIP_TAG
+RUNNING_IN_DOCKER=1
+GIT_BRANCH=$GIT_BRANCH
+NEXT_VERSION=$NEXT_VERSION
+RELEASE_VERSION=$RELEASE_VERSION
+RELEASE_TAG=$RELEASE_TAG
+GIT_REF=$GIT_REF
+SPARK_PACKAGE_VERSION=$SPARK_PACKAGE_VERSION
+ASF_USERNAME=$ASF_USERNAME
+GIT_NAME=$GIT_NAME
+GIT_EMAIL=$GIT_EMAIL
+GPG_KEY=$GPG_KEY
+ASF_PASSWORD=$ASF_PASSWORD
+GPG_PASSPHRASE=$GPG_PASSPHRASE
+RELEASE_STEP=$RELEASE_STEP
+EOF
+
+JAVA_VOL=
+if [ -n "$JAVA" ]; then
+  echo "JAVA_HOME=/opt/spark-java" >> $ENVFILE
+  JAVA_VOL="--volume $JAVA:/opt/spark-java"
+fi
+
+# SPARK-24530: Sphinx must work with python 3 to generate doc correctly.
+echo "SPHINXPYTHON=/opt/p35/bin/python" >> $ENVFILE
+
+echo "Building $RELEASE_TAG; output will be at $WORKDIR/output"
+docker run -ti \
+  --env-file "$ENVFILE" \
+  --volume "$WORKDIR:/opt/spark-rm" \
+  $JAVA_VOL \
+  "spark-rm:$IMGTAG"
diff --git a/dev/create-release/do-release.sh b/dev/create-release/do-release.sh
new file mode 100755
index 0000000000000..f1d4f3ab5ddec
--- /dev/null
+++ b/dev/create-release/do-release.sh
@@ -0,0 +1,81 @@
+#!/usr/bin/env bash
+
+#
+# Licensed to the Apache Software Foundation (ASF) under one or more
+# contributor license agreements.  See the NOTICE file distributed with
+# this work for additional information regarding copyright ownership.
+# The ASF licenses this file to You under the Apache License, Version 2.0
+# (the "License"); you may not use this file except in compliance with
+# the License.  You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+SELF=$(cd $(dirname $0) && pwd)
+. "$SELF/release-util.sh"
+
+while getopts "bn" opt; do
+  case $opt in
+    b) GIT_BRANCH=$OPTARG ;;
+    n) DRY_RUN=1 ;;
+    ?) error "Invalid option: $OPTARG" ;;
+  esac
+done
+
+if [ "$RUNNING_IN_DOCKER" = "1" ]; then
+  # Inside docker, need to import the GPG key stored in the current directory.
+  echo $GPG_PASSPHRASE | $GPG --passphrase-fd 0 --import "$SELF/gpg.key"
+
+  # We may need to adjust the path since JAVA_HOME may be overridden by the driver script.
+  if [ -n "$JAVA_HOME" ]; then
+    export PATH="$JAVA_HOME/bin:$PATH"
+  else
+    # JAVA_HOME for the openjdk package.
+    export JAVA_HOME=/usr
+  fi
+else
+  # Outside docker, need to ask for information about the release.
+  get_release_info
+fi
+
+function should_build {
+  local WHAT=$1
+  [ -z "$RELEASE_STEP" ] || [ "$WHAT" = "$RELEASE_STEP" ]
+}
+
+if should_build "tag" && [ $SKIP_TAG = 0 ]; then
+  run_silent "Creating release tag $RELEASE_TAG..." "tag.log" \
+    "$SELF/release-tag.sh"
+  echo "It may take some time for the tag to be synchronized to github."
+  echo "Press enter when you've verified that the new tag ($RELEASE_TAG) is available."
+  read
+else
+  echo "Skipping tag creation for $RELEASE_TAG."
+fi
+
+if should_build "build"; then
+  run_silent "Building Spark..." "build.log" \
+    "$SELF/release-build.sh" package
+else
+  echo "Skipping build step."
+fi
+
+if should_build "docs"; then
+  run_silent "Building documentation..." "docs.log" \
+    "$SELF/release-build.sh" docs
+else
+  echo "Skipping docs step."
+fi
+
+if should_build "publish"; then
+  run_silent "Publishing release" "publish.log" \
+    "$SELF/release-build.sh" publish-release
+else
+  echo "Skipping publish step."
+fi
diff --git a/dev/create-release/release-build.sh b/dev/create-release/release-build.sh
index 4e4b00c1a1159..320e1cd4674b0 100755
--- a/dev/create-release/release-build.sh
+++ b/dev/create-release/release-build.sh
@@ -17,6 +17,9 @@
 # limitations under the License.
 #
 
+SELF=$(cd $(dirname $0) && pwd)
+. "$SELF/release-util.sh"
+
 function exit_with_usage {
   cat << EOF
 usage: release-build.sh <package|docs|publish-snapshot|publish-release>
@@ -73,71 +76,78 @@ for env in ASF_USERNAME GPG_PASSPHRASE GPG_KEY; do
   fi
 done
 
-# Explicitly set locale in order to make `sort` output consistent across machines.
-# See https://stackoverflow.com/questions/28881 for more details.
-export LC_ALL=C
+export LC_ALL=C.UTF-8
+export LANG=C.UTF-8
 
 # Commit ref to checkout when building
 GIT_REF=${GIT_REF:-master}
 
 RELEASE_STAGING_LOCATION="https://dist.apache.org/repos/dist/dev/spark"
 
-GPG="gpg -u $GPG_KEY --no-tty --batch"
+GPG="gpg -u $GPG_KEY --no-tty --batch --pinentry-mode loopback"
 NEXUS_ROOT=https://repository.apache.org/service/local/staging
 NEXUS_PROFILE=d63f592e7eac0 # Profile for Spark staging uploads
 BASE_DIR=$(pwd)
 
-MVN="build/mvn --force"
-
-# Hive-specific profiles for some builds
-HIVE_PROFILES="-Phive -Phive-thriftserver"
-# Profiles for publishing snapshots and release to Maven Central
-PUBLISH_PROFILES="-Pmesos -Pyarn -Pkubernetes -Pflume $HIVE_PROFILES -Pspark-ganglia-lgpl -Pkinesis-asl"
-# Profiles for building binary releases
-BASE_RELEASE_PROFILES="-Pmesos -Pyarn -Pkubernetes -Pflume -Psparkr"
-# Scala 2.11 only profiles for some builds
-SCALA_2_11_PROFILES="-Pkafka-0-8"
-# Scala 2.12 only profiles for some builds
-SCALA_2_12_PROFILES="-Pscala-2.12"
+init_java
+init_maven_sbt
 
 rm -rf spark
-git clone https://gitbox.apache.org/repos/asf/spark.git
+git clone "$ASF_REPO"
 cd spark
 git checkout $GIT_REF
 git_hash=`git rev-parse --short HEAD`
 echo "Checked out Spark git hash $git_hash"
 
 if [ -z "$SPARK_VERSION" ]; then
-  SPARK_VERSION=$($MVN help:evaluate -Dexpression=project.version \
-    | grep -v INFO | grep -v WARNING | grep -v Download)
+  # Run $MVN in a separate command so that 'set -e' does the right thing.
+  TMP=$(mktemp)
+  $MVN help:evaluate -Dexpression=project.version > $TMP
+  SPARK_VERSION=$(cat $TMP | grep -v INFO | grep -v WARNING | grep -v Download)
+  rm $TMP
 fi
 
-# Verify we have the right java version set
-if [ -z "$JAVA_HOME" ]; then
-  echo "Please set JAVA_HOME."
-  exit 1
+# Depending on the version being built, certain extra profiles need to be activated, and
+# different versions of Scala are supported.
+BASE_PROFILES="-Pmesos -Pyarn"
+if [[ $SPARK_VERSION > "2.3" ]]; then
+  BASE_PROFILES="$BASE_PROFILES -Pkubernetes"
 fi
 
-java_version=$("${JAVA_HOME}"/bin/javac -version 2>&1 | cut -d " " -f 2)
+# TODO: revisit for Scala 2.13
 
-if [[ ! $SPARK_VERSION < "2.2." ]]; then
-  if [[ $java_version < "1.8." ]]; then
-    echo "Java version $java_version is less than required 1.8 for 2.2+"
-    echo "Please set JAVA_HOME correctly."
-    exit 1
-  fi
-else
-  if [[ $java_version > "1.7." ]]; then
-    if [ -z "$JAVA_7_HOME" ]; then
-      echo "Java version $java_version is higher than required 1.7 for pre-2.2"
-      echo "Please set JAVA_HOME correctly."
-      exit 1
-    else
-      export JAVA_HOME="$JAVA_7_HOME"
-    fi
+PUBLISH_SCALA_2_11=1
+SCALA_2_11_PROFILES="-Pscala-2.11"
+if [[ $SPARK_VERSION > "2.3" ]]; then
+  if [[ $SPARK_VERSION < "3.0." ]]; then
+    SCALA_2_11_PROFILES="-Pkafka-0-8 -Pflume $SCALA_2_11_PROFILES"
+  else
+    PUBLISH_SCALA_2_11=0
   fi
 fi
 
+PUBLISH_SCALA_2_12=0
+SCALA_2_12_PROFILES="-Pscala-2.12"
+if [[ $SPARK_VERSION < "3.0." ]]; then
+  SCALA_2_12_PROFILES="-Pscala-2.12 -Pflume"
+fi
+if [[ $SPARK_VERSION > "2.4" ]]; then
+  PUBLISH_SCALA_2_12=1
+fi
+
+# Hive-specific profiles for some builds
+HIVE_PROFILES="-Phive -Phive-thriftserver"
+# Profiles for publishing snapshots and release to Maven Central
+PUBLISH_PROFILES="$BASE_PROFILES $HIVE_PROFILES -Pspark-ganglia-lgpl -Pkinesis-asl"
+# Profiles for building binary releases
+BASE_RELEASE_PROFILES="$BASE_PROFILES -Psparkr"
+
+if [[ $JAVA_VERSION < "1.8." ]]; then
+  echo "Java version $JAVA_VERSION is less than required 1.8 for 2.2+"
+  echo "Please set JAVA_HOME correctly."
+  exit 1
+fi
+
 # This is a band-aid fix to avoid the failure of Maven nightly snapshot in some Jenkins
 # machines by explicitly calling /usr/sbin/lsof. Please see SPARK-22377 and the discussion
 # in its pull request.
@@ -161,6 +171,14 @@ if [[ "$1" == "package" ]]; then
   # Source and binary tarballs
   echo "Packaging release source tarballs"
   cp -r spark spark-$SPARK_VERSION
+
+  # For source release in v2.4+, exclude copy of binary license/notice
+  if [[ $SPARK_VERSION > "2.4" ]]; then
+    rm spark-$SPARK_VERSION/LICENSE-binary
+    rm spark-$SPARK_VERSION/NOTICE-binary
+    rm -r spark-$SPARK_VERSION/licenses-binary
+  fi
+
   tar cvzf spark-$SPARK_VERSION.tgz spark-$SPARK_VERSION
   echo $GPG_PASSPHRASE | $GPG --passphrase-fd 0 --armour --output spark-$SPARK_VERSION.tgz.asc \
     --detach-sig spark-$SPARK_VERSION.tgz
@@ -168,20 +186,34 @@ if [[ "$1" == "package" ]]; then
     SHA512 spark-$SPARK_VERSION.tgz > spark-$SPARK_VERSION.tgz.sha512
   rm -rf spark-$SPARK_VERSION
 
+  ZINC_PORT=3035
+
   # Updated for each binary build
   make_binary_release() {
     NAME=$1
-    FLAGS=$2
-    ZINC_PORT=$3
-    BUILD_PACKAGE=$4
-    cp -r spark spark-$SPARK_VERSION-bin-$NAME
+    FLAGS="$MVN_EXTRA_OPTS -B $BASE_RELEASE_PROFILES $2"
+    # BUILD_PACKAGE can be "withpip", "withr", or both as "withpip,withr"
+    BUILD_PACKAGE=$3
+    SCALA_VERSION=$4
+
+    PIP_FLAG=""
+    if [[ $BUILD_PACKAGE == *"withpip"* ]]; then
+      PIP_FLAG="--pip"
+    fi
+    R_FLAG=""
+    if [[ $BUILD_PACKAGE == *"withr"* ]]; then
+      R_FLAG="--r"
+    fi
+
+    # We increment the Zinc port each time to avoid OOM's and other craziness if multiple builds
+    # share the same Zinc server.
+    ZINC_PORT=$((ZINC_PORT + 1))
 
+    echo "Building binary dist $NAME"
+    cp -r spark spark-$SPARK_VERSION-bin-$NAME
     cd spark-$SPARK_VERSION-bin-$NAME
 
-    # TODO There should probably be a flag to make-distribution to allow 2.12 support
-    #if [[ $FLAGS == *scala-2.12* ]]; then
-    #  ./dev/change-scala-version.sh 2.12
-    #fi
+    ./dev/change-scala-version.sh $SCALA_VERSION
 
     export ZINC_PORT=$ZINC_PORT
     echo "Creating distribution: $NAME ($FLAGS)"
@@ -194,18 +226,13 @@ if [[ "$1" == "package" ]]; then
     # Get maven home set by MVN
     MVN_HOME=`$MVN -version 2>&1 | grep 'Maven home' | awk '{print $NF}'`
 
+    echo "Creating distribution"
+    ./dev/make-distribution.sh --name $NAME --mvn $MVN_HOME/bin/mvn --tgz \
+      $PIP_FLAG $R_FLAG $FLAGS \
+      -DzincPort=$ZINC_PORT 2>&1 >  ../binary-release-$NAME.log
+    cd ..
 
-    if [ -z "$BUILD_PACKAGE" ]; then
-      echo "Creating distribution without PIP/R package"
-      ./dev/make-distribution.sh --name $NAME --mvn $MVN_HOME/bin/mvn --tgz $FLAGS \
-        -DzincPort=$ZINC_PORT 2>&1 >  ../binary-release-$NAME.log
-      cd ..
-    elif [[ "$BUILD_PACKAGE" == "withr" ]]; then
-      echo "Creating distribution with R package"
-      ./dev/make-distribution.sh --name $NAME --mvn $MVN_HOME/bin/mvn --tgz --r $FLAGS \
-        -DzincPort=$ZINC_PORT 2>&1 >  ../binary-release-$NAME.log
-      cd ..
-
+    if [[ -n $R_FLAG ]]; then
       echo "Copying and signing R source package"
       R_DIST_NAME=SparkR_$SPARK_VERSION.tar.gz
       cp spark-$SPARK_VERSION-bin-$NAME/R/$R_DIST_NAME .
@@ -216,12 +243,9 @@ if [[ "$1" == "package" ]]; then
       echo $GPG_PASSPHRASE | $GPG --passphrase-fd 0 --print-md \
         SHA512 $R_DIST_NAME > \
         $R_DIST_NAME.sha512
-    else
-      echo "Creating distribution with PIP package"
-      ./dev/make-distribution.sh --name $NAME --mvn $MVN_HOME/bin/mvn --tgz --pip $FLAGS \
-        -DzincPort=$ZINC_PORT 2>&1 >  ../binary-release-$NAME.log
-      cd ..
+    fi
 
+    if [[ -n $PIP_FLAG ]]; then
       echo "Copying and signing python distribution"
       PYTHON_DIST_NAME=pyspark-$PYSPARK_VERSION.tar.gz
       cp spark-$SPARK_VERSION-bin-$NAME/python/dist/$PYTHON_DIST_NAME .
@@ -244,18 +268,50 @@ if [[ "$1" == "package" ]]; then
       spark-$SPARK_VERSION-bin-$NAME.tgz.sha512
   }
 
-  # TODO: Check exit codes of children here:
-  # http://stackoverflow.com/questions/1570262/shell-get-exit-code-of-background-process
+  # List of binary packages built. Populates two associative arrays, where the key is the "name" of
+  # the package being built, and the values are respectively the needed maven arguments for building
+  # the package, and any extra package needed for that particular combination.
+  #
+  # In dry run mode, only build the first one. The keys in BINARY_PKGS_ARGS are used as the
+  # list of packages to be built, so it's ok for things to be missing in BINARY_PKGS_EXTRA.
+
+  declare -A BINARY_PKGS_ARGS
+  BINARY_PKGS_ARGS["hadoop2.7"]="-Phadoop-2.7 $HIVE_PROFILES"
+  BINARY_PKGS_ARGS["without-hadoop"]="-Phadoop-provided"
+  if [[ $SPARK_VERSION < "3.0." ]]; then
+    BINARY_PKGS_ARGS["hadoop2.6"]="-Phadoop-2.6 $HIVE_PROFILES"
+  fi
+
+  declare -A BINARY_PKGS_EXTRA
+  BINARY_PKGS_EXTRA["hadoop2.7"]="withpip,withr"
+
+  if [[ $PUBLISH_SCALA_2_11 = 1 ]]; then
+    echo "Packages to build: ${!BINARY_PKGS_ARGS[@]}"
+    for key in ${!BINARY_PKGS_ARGS[@]}; do
+      args=${BINARY_PKGS_ARGS[$key]}
+      extra=${BINARY_PKGS_EXTRA[$key]}
+      if ! make_binary_release "$key" "$SCALA_2_11_PROFILES $args" "$extra" "2.11"; then
+        error "Failed to build $key package. Check logs for details."
+      fi
+    done
+  fi
+
+  if [[ $PUBLISH_SCALA_2_12 = 1 ]]; then
+    echo "Packages to build: ${!BINARY_PKGS_ARGS[@]}"
+    for key in ${!BINARY_PKGS_ARGS[@]}; do
+      args=${BINARY_PKGS_ARGS[$key]}
+      extra=${BINARY_PKGS_EXTRA[$key]}
+      if ! make_binary_release "$key" "$SCALA_2_12_PROFILES $args" "$extra" "2.12"; then
+        error "Failed to build $key package. Check logs for details."
+      fi
+    done
+  fi
 
-  # We increment the Zinc port each time to avoid OOM's and other craziness if multiple builds
-  # share the same Zinc server.
-  make_binary_release "hadoop2.6" "-Phadoop-2.6 $HIVE_PROFILES $SCALA_2_11_PROFILES $BASE_RELEASE_PROFILES" "3035" "withr" &
-  make_binary_release "hadoop2.7" "-Phadoop-2.7 $HIVE_PROFILES $SCALA_2_11_PROFILES $BASE_RELEASE_PROFILES" "3036" "withpip" &
-  make_binary_release "without-hadoop" "-Phadoop-provided $SCALA_2_11_PROFILES $BASE_RELEASE_PROFILES" "3038" &
-  wait
   rm -rf spark-$SPARK_VERSION-bin-*/
 
-  svn co --depth=empty $RELEASE_STAGING_LOCATION svn-spark
+  if ! is_dry_run; then
+    svn co --depth=empty $RELEASE_STAGING_LOCATION svn-spark
+  fi
   rm -rf "svn-spark/${DEST_DIR_NAME}-bin"
   mkdir -p "svn-spark/${DEST_DIR_NAME}-bin"
 
@@ -263,12 +319,15 @@ if [[ "$1" == "package" ]]; then
   cp spark-* "svn-spark/${DEST_DIR_NAME}-bin/"
   cp pyspark-* "svn-spark/${DEST_DIR_NAME}-bin/"
   cp SparkR_* "svn-spark/${DEST_DIR_NAME}-bin/"
-  svn add "svn-spark/${DEST_DIR_NAME}-bin"
+  if ! is_dry_run; then
+    svn add "svn-spark/${DEST_DIR_NAME}-bin"
+
+    cd svn-spark
+    svn ci --username $ASF_USERNAME --password "$ASF_PASSWORD" -m"Apache Spark $SPARK_PACKAGE_VERSION" --no-auth-cache
+    cd ..
+    rm -rf svn-spark
+  fi
 
-  cd svn-spark
-  svn ci --username $ASF_USERNAME --password "$ASF_PASSWORD" -m"Apache Spark $SPARK_PACKAGE_VERSION"
-  cd ..
-  rm -rf svn-spark
   exit 0
 fi
 
@@ -282,18 +341,24 @@ if [[ "$1" == "docs" ]]; then
   cd ..
   cd ..
 
-  svn co --depth=empty $RELEASE_STAGING_LOCATION svn-spark
+  if ! is_dry_run; then
+    svn co --depth=empty $RELEASE_STAGING_LOCATION svn-spark
+  fi
   rm -rf "svn-spark/${DEST_DIR_NAME}-docs"
   mkdir -p "svn-spark/${DEST_DIR_NAME}-docs"
 
   echo "Copying release documentation"
   cp -R "spark/docs/_site" "svn-spark/${DEST_DIR_NAME}-docs/"
-  svn add "svn-spark/${DEST_DIR_NAME}-docs"
+  if ! is_dry_run; then
+    svn add "svn-spark/${DEST_DIR_NAME}-docs"
 
-  cd svn-spark
-  svn ci --username $ASF_USERNAME --password "$ASF_PASSWORD" -m"Apache Spark $SPARK_PACKAGE_VERSION docs"
-  cd ..
-  rm -rf svn-spark
+    cd svn-spark
+    svn ci --username $ASF_USERNAME --password "$ASF_PASSWORD" -m"Apache Spark $SPARK_PACKAGE_VERSION docs" --no-auth-cache
+    cd ..
+    rm -rf svn-spark
+  fi
+
+  mv "spark/docs/_site" docs/
   exit 0
 fi
 
@@ -318,10 +383,7 @@ if [[ "$1" == "publish-snapshot" ]]; then
   # Generate random point for Zinc
   export ZINC_PORT=$(python -S -c "import random; print random.randrange(3030,4030)")
 
-  $MVN -DzincPort=$ZINC_PORT --settings $tmp_settings -DskipTests $SCALA_2_11_PROFILES $PUBLISH_PROFILES deploy
-  #./dev/change-scala-version.sh 2.12
-  #$MVN -DzincPort=$ZINC_PORT --settings $tmp_settings \
-  #  -DskipTests $SCALA_2_12_PROFILES $PUBLISH_PROFILES clean deploy
+  $MVN -DzincPort=$ZINC_PORT --settings $tmp_settings -DskipTests $SCALA_2_12_PROFILES $PUBLISH_PROFILES deploy
 
   rm $tmp_settings
   cd ..
@@ -338,26 +400,36 @@ if [[ "$1" == "publish-release" ]]; then
 
   # Using Nexus API documented here:
   # https://support.sonatype.com/entries/39720203-Uploading-to-a-Staging-Repository-via-REST-API
-  echo "Creating Nexus staging repository"
-  repo_request="<promoteRequest><data><description>Apache Spark $SPARK_VERSION (commit $git_hash)</description></data></promoteRequest>"
-  out=$(curl -X POST -d "$repo_request" -u $ASF_USERNAME:$ASF_PASSWORD \
-    -H "Content-Type:application/xml" -v \
-    $NEXUS_ROOT/profiles/$NEXUS_PROFILE/start)
-  staged_repo_id=$(echo $out | sed -e "s/.*\(orgapachespark-[0-9]\{4\}\).*/\1/")
-  echo "Created Nexus staging repository: $staged_repo_id"
+  if ! is_dry_run; then
+    echo "Creating Nexus staging repository"
+    repo_request="<promoteRequest><data><description>Apache Spark $SPARK_VERSION (commit $git_hash)</description></data></promoteRequest>"
+    out=$(curl -X POST -d "$repo_request" -u $ASF_USERNAME:$ASF_PASSWORD \
+      -H "Content-Type:application/xml" -v \
+      $NEXUS_ROOT/profiles/$NEXUS_PROFILE/start)
+    staged_repo_id=$(echo $out | sed -e "s/.*\(orgapachespark-[0-9]\{4\}\).*/\1/")
+    echo "Created Nexus staging repository: $staged_repo_id"
+  fi
 
   tmp_repo=$(mktemp -d spark-repo-XXXXX)
 
   # Generate random point for Zinc
   export ZINC_PORT=$(python -S -c "import random; print random.randrange(3030,4030)")
 
-  $MVN -DzincPort=$ZINC_PORT -Dmaven.repo.local=$tmp_repo -DskipTests $SCALA_2_11_PROFILES $PUBLISH_PROFILES clean install
+  # TODO: revisit for Scala 2.13 support
 
-  #./dev/change-scala-version.sh 2.12
-  #$MVN -DzincPort=$ZINC_PORT -Dmaven.repo.local=$tmp_repo \
-  #  -DskipTests $SCALA_2_12_PROFILES §$PUBLISH_PROFILES clean install
+#  if ! is_dry_run && [[ $PUBLISH_SCALA_2_11 = 1 ]]; then
+  if [[ $PUBLISH_SCALA_2_11 = 1 ]]; then
+    ./dev/change-scala-version.sh 2.11
+    $MVN -DzincPort=$ZINC_PORT -Dmaven.repo.local=$tmp_repo -DskipTests \
+      $SCALA_2_11_PROFILES $PUBLISH_PROFILES clean install
+  fi
 
-  #./dev/change-scala-version.sh 2.11
+#  if ! is_dry_run && [[ $PUBLISH_SCALA_2_12 = 1 ]]; then
+  if [[ $PUBLISH_SCALA_2_12 = 1 ]]; then
+    ./dev/change-scala-version.sh 2.12
+    $MVN -DzincPort=$((ZINC_PORT + 2)) -Dmaven.repo.local=$tmp_repo -DskipTests \
+      $SCALA_2_11_PROFILES $PUBLISH_PROFILES clean install
+  fi
 
   pushd $tmp_repo/org/apache/spark
 
@@ -380,23 +452,26 @@ if [[ "$1" == "publish-release" ]]; then
     sha1sum $file | cut -f1 -d' ' > $file.sha1
   done
 
-  nexus_upload=$NEXUS_ROOT/deployByRepositoryId/$staged_repo_id
-  echo "Uplading files to $nexus_upload"
-  for file in $(find . -type f)
-  do
-    # strip leading ./
-    file_short=$(echo $file | sed -e "s/\.\///")
-    dest_url="$nexus_upload/org/apache/spark/$file_short"
-    echo "  Uploading $file_short"
-    curl -u $ASF_USERNAME:$ASF_PASSWORD --upload-file $file_short $dest_url
-  done
+  if ! is_dry_run; then
+    nexus_upload=$NEXUS_ROOT/deployByRepositoryId/$staged_repo_id
+    echo "Uplading files to $nexus_upload"
+    for file in $(find . -type f)
+    do
+      # strip leading ./
+      file_short=$(echo $file | sed -e "s/\.\///")
+      dest_url="$nexus_upload/org/apache/spark/$file_short"
+      echo "  Uploading $file_short"
+      curl -u $ASF_USERNAME:$ASF_PASSWORD --upload-file $file_short $dest_url
+    done
+
+    echo "Closing nexus staging repository"
+    repo_request="<promoteRequest><data><stagedRepositoryId>$staged_repo_id</stagedRepositoryId><description>Apache Spark $SPARK_VERSION (commit $git_hash)</description></data></promoteRequest>"
+    out=$(curl -X POST -d "$repo_request" -u $ASF_USERNAME:$ASF_PASSWORD \
+      -H "Content-Type:application/xml" -v \
+      $NEXUS_ROOT/profiles/$NEXUS_PROFILE/finish)
+    echo "Closed Nexus staging repository: $staged_repo_id"
+  fi
 
-  echo "Closing nexus staging repository"
-  repo_request="<promoteRequest><data><stagedRepositoryId>$staged_repo_id</stagedRepositoryId><description>Apache Spark $SPARK_VERSION (commit $git_hash)</description></data></promoteRequest>"
-  out=$(curl -X POST -d "$repo_request" -u $ASF_USERNAME:$ASF_PASSWORD \
-    -H "Content-Type:application/xml" -v \
-    $NEXUS_ROOT/profiles/$NEXUS_PROFILE/finish)
-  echo "Closed Nexus staging repository: $staged_repo_id"
   popd
   rm -rf $tmp_repo
   cd ..
diff --git a/dev/create-release/release-tag.sh b/dev/create-release/release-tag.sh
index 319aa069a096b..8024440759eb5 100755
--- a/dev/create-release/release-tag.sh
+++ b/dev/create-release/release-tag.sh
@@ -17,6 +17,9 @@
 # limitations under the License.
 #
 
+SELF=$(cd $(dirname $0) && pwd)
+. "$SELF/release-util.sh"
+
 function exit_with_usage {
   local NAME=$(basename $0)
   cat << EOF
@@ -37,6 +40,7 @@ EOF
 }
 
 set -e
+set -o pipefail
 
 if [[ $@ == *"help"* ]]; then
   exit_with_usage
@@ -55,8 +59,10 @@ for env in ASF_USERNAME ASF_PASSWORD RELEASE_VERSION RELEASE_TAG NEXT_VERSION GI
   fi
 done
 
+init_java
+init_maven_sbt
+
 ASF_SPARK_REPO="gitbox.apache.org/repos/asf/spark.git"
-MVN="build/mvn --force"
 
 rm -rf spark
 git clone "https://$ASF_USERNAME:$ASF_PASSWORD@$ASF_SPARK_REPO" -b $GIT_BRANCH
@@ -95,9 +101,15 @@ sed -i".tmp7" 's/SPARK_VERSION_SHORT:.*$/SPARK_VERSION_SHORT: '"$R_NEXT_VERSION"
 
 git commit -a -m "Preparing development version $NEXT_VERSION"
 
-# Push changes
-git push origin $RELEASE_TAG
-git push origin HEAD:$GIT_BRANCH
-
-cd ..
-rm -rf spark
+if ! is_dry_run; then
+  # Push changes
+  git push origin $RELEASE_TAG
+  git push origin HEAD:$GIT_BRANCH
+
+  cd ..
+  rm -rf spark
+else
+  cd ..
+  mv spark spark.tag
+  echo "Clone with version changes and tag available as spark.tag in the output directory."
+fi
diff --git a/dev/create-release/release-util.sh b/dev/create-release/release-util.sh
new file mode 100755
index 0000000000000..1282e15e79913
--- /dev/null
+++ b/dev/create-release/release-util.sh
@@ -0,0 +1,232 @@
+#!/usr/bin/env bash
+
+#
+# Licensed to the Apache Software Foundation (ASF) under one or more
+# contributor license agreements.  See the NOTICE file distributed with
+# this work for additional information regarding copyright ownership.
+# The ASF licenses this file to You under the Apache License, Version 2.0
+# (the "License"); you may not use this file except in compliance with
+# the License.  You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+DRY_RUN=${DRY_RUN:-0}
+GPG="gpg --no-tty --batch"
+ASF_REPO="https://gitbox.apache.org/repos/asf/spark.git"
+ASF_REPO_WEBUI="https://gitbox.apache.org/repos/asf?p=spark.git"
+ASF_GITHUB_REPO="https://github.com/apache/spark"
+
+function error {
+  echo "$*"
+  exit 1
+}
+
+function read_config {
+  local PROMPT="$1"
+  local DEFAULT="$2"
+  local REPLY=
+
+  read -p "$PROMPT [$DEFAULT]: " REPLY
+  local RETVAL="${REPLY:-$DEFAULT}"
+  if [ -z "$RETVAL" ]; then
+    error "$PROMPT is must be provided."
+  fi
+  echo "$RETVAL"
+}
+
+function parse_version {
+  grep -e '<version>.*</version>' | \
+    head -n 2 | tail -n 1 | cut -d'>' -f2 | cut -d '<' -f1
+}
+
+function run_silent {
+  local BANNER="$1"
+  local LOG_FILE="$2"
+  shift 2
+
+  echo "========================"
+  echo "= $BANNER"
+  echo "Command: $@"
+  echo "Log file: $LOG_FILE"
+
+  "$@" 1>"$LOG_FILE" 2>&1
+
+  local EC=$?
+  if [ $EC != 0 ]; then
+    echo "Command FAILED. Check full logs for details."
+    tail "$LOG_FILE"
+    exit $EC
+  fi
+}
+
+function fcreate_secure {
+  local FPATH="$1"
+  rm -f "$FPATH"
+  touch "$FPATH"
+  chmod 600 "$FPATH"
+}
+
+function check_for_tag {
+  curl -s --head --fail "$ASF_GITHUB_REPO/releases/tag/$1" > /dev/null
+}
+
+function get_release_info {
+  if [ -z "$GIT_BRANCH" ]; then
+    # If no branch is specified, found out the latest branch from the repo.
+    GIT_BRANCH=$(git ls-remote --heads "$ASF_REPO" |
+      grep -v refs/heads/master |
+      awk '{print $2}' |
+      sort -r |
+      head -n 1 |
+      cut -d/ -f3)
+  fi
+
+  export GIT_BRANCH=$(read_config "Branch" "$GIT_BRANCH")
+
+  # Find the current version for the branch.
+  local VERSION=$(curl -s "$ASF_REPO_WEBUI;a=blob_plain;f=pom.xml;hb=refs/heads/$GIT_BRANCH" |
+    parse_version)
+  echo "Current branch version is $VERSION."
+
+  if [[ ! $VERSION =~ .*-SNAPSHOT ]]; then
+    error "Not a SNAPSHOT version: $VERSION"
+  fi
+
+  NEXT_VERSION="$VERSION"
+  RELEASE_VERSION="${VERSION/-SNAPSHOT/}"
+  SHORT_VERSION=$(echo "$VERSION" | cut -d . -f 1-2)
+  local REV=$(echo "$RELEASE_VERSION" | cut -d . -f 3)
+
+  # Find out what rc is being prepared.
+  # - If the current version is "x.y.0", then this is rc1 of the "x.y.0" release.
+  # - If not, need to check whether the previous version has been already released or not.
+  #   - If it has, then we're building rc1 of the current version.
+  #   - If it has not, we're building the next RC of the previous version.
+  local RC_COUNT
+  if [ $REV != 0 ]; then
+    local PREV_REL_REV=$((REV - 1))
+    local PREV_REL_TAG="v${SHORT_VERSION}.${PREV_REL_REV}"
+    if check_for_tag "$PREV_REL_TAG"; then
+      RC_COUNT=1
+      REV=$((REV + 1))
+      NEXT_VERSION="${SHORT_VERSION}.${REV}-SNAPSHOT"
+    else
+      RELEASE_VERSION="${SHORT_VERSION}.${PREV_REL_REV}"
+      RC_COUNT=$(git ls-remote --tags "$ASF_REPO" "v${RELEASE_VERSION}-rc*" | wc -l)
+      RC_COUNT=$((RC_COUNT + 1))
+    fi
+  else
+    REV=$((REV + 1))
+    NEXT_VERSION="${SHORT_VERSION}.${REV}-SNAPSHOT"
+    RC_COUNT=1
+  fi
+
+  export NEXT_VERSION
+  export RELEASE_VERSION=$(read_config "Release" "$RELEASE_VERSION")
+
+  RC_COUNT=$(read_config "RC #" "$RC_COUNT")
+
+  # Check if the RC already exists, and if re-creating the RC, skip tag creation.
+  RELEASE_TAG="v${RELEASE_VERSION}-rc${RC_COUNT}"
+  SKIP_TAG=0
+  if check_for_tag "$RELEASE_TAG"; then
+    read -p "$RELEASE_TAG already exists. Continue anyway [y/n]? " ANSWER
+    if [ "$ANSWER" != "y" ]; then
+      error "Exiting."
+    fi
+    SKIP_TAG=1
+  fi
+
+
+  export RELEASE_TAG
+
+  GIT_REF="$RELEASE_TAG"
+  if is_dry_run; then
+    echo "This is a dry run. Please confirm the ref that will be built for testing."
+    if [[ $SKIP_TAG = 0 ]]; then
+      GIT_REF="$GIT_BRANCH"
+    fi
+    GIT_REF=$(read_config "Ref" "$GIT_REF")
+  fi
+  export GIT_REF
+  export SPARK_PACKAGE_VERSION="$RELEASE_TAG"
+
+  # Gather some user information.
+  export ASF_USERNAME=$(read_config "ASF user" "$LOGNAME")
+
+  GIT_NAME=$(git config user.name || echo "")
+  export GIT_NAME=$(read_config "Full name" "$GIT_NAME")
+
+  export GIT_EMAIL="$ASF_USERNAME@apache.org"
+  export GPG_KEY=$(read_config "GPG key" "$GIT_EMAIL")
+
+  cat <<EOF
+================
+Release details:
+BRANCH:     $GIT_BRANCH
+VERSION:    $RELEASE_VERSION
+TAG:        $RELEASE_TAG
+NEXT:       $NEXT_VERSION
+
+ASF USER:   $ASF_USERNAME
+GPG KEY:    $GPG_KEY
+FULL NAME:  $GIT_NAME
+E-MAIL:     $GIT_EMAIL
+================
+EOF
+
+  read -p "Is this info correct [y/n]? " ANSWER
+  if [ "$ANSWER" != "y" ]; then
+    echo "Exiting."
+    exit 1
+  fi
+
+  if ! is_dry_run; then
+    if [ -z "$ASF_PASSWORD" ]; then
+      stty -echo && printf "ASF password: " && read ASF_PASSWORD && printf '\n' && stty echo
+    fi
+  else
+    ASF_PASSWORD="***INVALID***"
+  fi
+
+  if [ -z "$GPG_PASSPHRASE" ]; then
+    stty -echo && printf "GPG passphrase: " && read GPG_PASSPHRASE && printf '\n' && stty echo
+  fi
+
+  export ASF_PASSWORD
+  export GPG_PASSPHRASE
+}
+
+function is_dry_run {
+  [[ $DRY_RUN = 1 ]]
+}
+
+# Initializes JAVA_VERSION to the version of the JVM in use.
+function init_java {
+  if [ -z "$JAVA_HOME" ]; then
+    error "JAVA_HOME is not set."
+  fi
+  JAVA_VERSION=$("${JAVA_HOME}"/bin/javac -version 2>&1 | cut -d " " -f 2)
+  export JAVA_VERSION
+}
+
+# Initializes MVN_EXTRA_OPTS and SBT_OPTS depending on the JAVA_VERSION in use. Requires init_java.
+function init_maven_sbt {
+  MVN="build/mvn -B"
+  MVN_EXTRA_OPTS=
+  SBT_OPTS=
+  if [[ $JAVA_VERSION < "1.8." ]]; then
+    # Needed for maven central when using Java 7.
+    SBT_OPTS="-Dhttps.protocols=TLSv1.1,TLSv1.2"
+    MVN_EXTRA_OPTS="-Dhttps.protocols=TLSv1.1,TLSv1.2"
+    MVN="$MVN $MVN_EXTRA_OPTS"
+  fi
+  export MVN MVN_EXTRA_OPTS SBT_OPTS
+}
diff --git a/dev/create-release/spark-rm/Dockerfile b/dev/create-release/spark-rm/Dockerfile
new file mode 100644
index 0000000000000..4bfecedbf0406
--- /dev/null
+++ b/dev/create-release/spark-rm/Dockerfile
@@ -0,0 +1,91 @@
+#
+# Licensed to the Apache Software Foundation (ASF) under one or more
+# contributor license agreements.  See the NOTICE file distributed with
+# this work for additional information regarding copyright ownership.
+# The ASF licenses this file to You under the Apache License, Version 2.0
+# (the "License"); you may not use this file except in compliance with
+# the License.  You may obtain a copy of the License at
+#
+#    http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+#
+
+# Image for building Spark releases. Based on Ubuntu 18.04.
+#
+# Includes:
+# * Java 8
+# * Ivy
+# * Python/PyPandoc (2.7.15/3.6.7)
+# * R-base/R-base-dev (3.6.1)
+# * Ruby 2.3 build utilities
+
+FROM ubuntu:18.04
+
+# For apt to be noninteractive
+ENV DEBIAN_FRONTEND noninteractive
+ENV DEBCONF_NONINTERACTIVE_SEEN true
+
+# These arguments are just for reuse and not really meant to be customized.
+ARG APT_INSTALL="apt-get install --no-install-recommends -y"
+
+ARG BASE_PIP_PKGS="setuptools wheel virtualenv"
+ARG PIP_PKGS="pyopenssl pypandoc numpy pygments sphinx"
+
+# Install extra needed repos and refresh.
+# - CRAN repo
+# - Ruby repo (for doc generation)
+#
+# This is all in a single "RUN" command so that if anything changes, "apt update" is run to fetch
+# the most current package versions (instead of potentially using old versions cached by docker).
+RUN apt-get clean && apt-get update && $APT_INSTALL gnupg ca-certificates && \
+  echo 'deb https://cloud.r-project.org/bin/linux/ubuntu bionic-cran35/' >> /etc/apt/sources.list && \
+  gpg --keyserver keyserver.ubuntu.com --recv-key E298A3A825C0D65DFD57CBB651716619E084DAB9 && \
+  gpg -a --export E084DAB9 | apt-key add - && \
+  apt-get clean && \
+  rm -rf /var/lib/apt/lists/* && \
+  apt-get clean && \
+  apt-get update && \
+  $APT_INSTALL software-properties-common && \
+  apt-add-repository -y ppa:brightbox/ruby-ng && \
+  apt-get update && \
+  # Install openjdk 8.
+  $APT_INSTALL openjdk-8-jdk && \
+  update-alternatives --set java /usr/lib/jvm/java-8-openjdk-amd64/jre/bin/java && \
+  # Install build / source control tools
+  $APT_INSTALL curl wget git maven ivy subversion make gcc lsof libffi-dev \
+    pandoc pandoc-citeproc libssl-dev libcurl4-openssl-dev libxml2-dev && \
+  curl -sL https://deb.nodesource.com/setup_11.x | bash && \
+  $APT_INSTALL nodejs && \
+  # Install needed python packages. Use pip for installing packages (for consistency).
+  $APT_INSTALL libpython2.7-dev libpython3-dev python-pip python3-pip && \
+  pip install $BASE_PIP_PKGS && \
+  pip install $PIP_PKGS && \
+  cd && \
+  virtualenv -p python3 /opt/p35 && \
+  . /opt/p35/bin/activate && \
+  pip install $BASE_PIP_PKGS && \
+  pip install $PIP_PKGS && \
+  # Install R packages and dependencies used when building.
+  # R depends on pandoc*, libssl (which are installed above).
+  $APT_INSTALL r-base r-base-dev && \
+  $APT_INSTALL texlive-latex-base texlive texlive-fonts-extra texinfo qpdf && \
+  Rscript -e "install.packages(c('curl', 'xml2', 'httr', 'devtools', 'testthat', 'knitr', 'rmarkdown', 'roxygen2', 'e1071', 'survival'), repos='https://cloud.r-project.org/')" && \
+  Rscript -e "devtools::install_github('jimhester/lintr')" && \
+  # Install tools needed to build the documentation.
+  $APT_INSTALL ruby2.3 ruby2.3-dev mkdocs && \
+  gem install jekyll --no-rdoc --no-ri -v 3.8.6 && \
+  gem install jekyll-redirect-from && \
+  gem install pygments.rb
+
+WORKDIR /opt/spark-rm/output
+
+ARG UID
+RUN useradd -m -s /bin/bash -p spark-rm -u $UID spark-rm
+USER spark-rm:spark-rm
+
+ENTRYPOINT [ "/opt/spark-rm/do-release.sh" ]