GH-27919: [CI][C++] Add a nightly job to test offline build

ci/scripts/cpp_build.sh

@@ -22,6 +22,7 @@ set -ex
 source_dir=${1}/cpp
 build_dir=${2}/cpp
 
+: ${ARROW_OFFLINE:=OFF}
 : ${ARROW_USE_CCACHE:=OFF}
 : ${BUILD_DOCS_CPP:=OFF}
 
@@ -98,6 +99,16 @@ esac
 mkdir -p ${build_dir}
 pushd ${build_dir}
 
+if [ "${ARROW_OFFLINE}" = "ON" ]; then
+  ${source_dir}/thirdparty/download_dependencies.sh ${PWD}/thirdparty > \
+    enable_offline_build.sh
+  . enable_offline_build.sh
+  # We can't use mv because we can't remove /etc/resolv.conf in Docker
+  # container.
+  cp /etc/resolv.conf{,.bak}
+  echo > /etc/resolv.conf
+fi
+
 if [ "${ARROW_EMSCRIPTEN:-OFF}" = "ON" ]; then
   if [ "${UBUNTU}" = "20.04" ]; then
     echo "arrow emscripten build is not supported on Ubuntu 20.04, run with UBUNTU=22.04"
@@ -237,6 +248,10 @@ time cmake --build . --target install
 # Save disk space by removing large temporary build products
 find . -name "*.o" -delete
 
+if [ "${ARROW_OFFLINE}" = "ON" ]; then
+  cp /etc/resolv.conf{.bak,}
+fi
+
 popd
 
 if [ -x "$(command -v ldconfig)" ]; then

dev/tasks/tasks.yml

@@ -1098,6 +1098,14 @@ tasks:
         UBUNTU: 20.04
       image: ubuntu-cpp-bundled
 
+  test-ubuntu-24.04-cpp-bundled-offline:
+    ci: github
+    template: docker-tests/github.linux.yml
+    params:
+      env:
+        UBUNTU: 24.04
+      image: ubuntu-cpp-bundled-offline
+
   test-ubuntu-24.04-cpp-gcc-13-bundled:
     ci: github
     template: docker-tests/github.linux.yml

docker-compose.yml

@@ -150,6 +150,7 @@ x-hierarchy:
     - ubuntu-r
     - ubuntu-r-only-r
   - ubuntu-cpp-bundled
+  - ubuntu-cpp-bundled-offline
   - ubuntu-cpp-minimal
   - ubuntu-cuda-cpp:
     - ubuntu-cuda-python
@@ -484,6 +485,30 @@ services:
     volumes: *ubuntu-volumes
     command: *cpp-command
 
+  ubuntu-cpp-bundled-offline:
+    # Arrow build with BUNDLED dependencies with downloaded dependencies.
+    image: ${REPO}:${ARCH}-ubuntu-${UBUNTU}-cpp-minimal
+    build:
+      context: .
+      dockerfile: ci/docker/ubuntu-${UBUNTU}-cpp-minimal.dockerfile
+      cache_from:
+        - ${REPO}:${ARCH}-ubuntu-${UBUNTU}-cpp-minimal
+      args:
+        arch: ${ARCH}
+        base: "${ARCH}/ubuntu:${UBUNTU}"
+        llvm: ${LLVM}
+    shm_size: *shm-size
+    ulimits: *ulimits
+    environment:
+      <<: [*common, *ccache, *sccache, *cpp]
+      ARROW_DEPENDENCY_SOURCE: BUNDLED
+      ARROW_OFFLINE: ON
+      # Apache ORC always uses external orc-format.
+      ARROW_ORC: OFF
+      CMAKE_GENERATOR: "Unix Makefiles"
+    volumes: *ubuntu-volumes
+    command: *cpp-command
+
   ubuntu-cpp-minimal:
     # Arrow build with minimal components/dependencies
     image: ${REPO}:${ARCH}-ubuntu-${UBUNTU}-cpp-minimal