diff --git a/BUILDING.txt b/BUILDING.txt index 5f40a0d7dc3..b872d7e4194 100644 --- a/BUILDING.txt +++ b/BUILDING.txt @@ -492,39 +492,66 @@ Building on CentOS 8 ---------------------------------------------------------------------------------- -Building on Windows +Building on Windows 10 ---------------------------------------------------------------------------------- Requirements: -* Windows System +* Windows 10 * JDK 1.8 -* Maven 3.0 or later -* Boost 1.72 -* Protocol Buffers 3.7.1 -* CMake 3.19 or newer -* Visual Studio 2010 Professional or Higher -* Windows SDK 8.1 (if building CPU rate control for the container executor) -* zlib headers (if building native code bindings for zlib) +* Maven 3.0 or later (maven.apache.org) +* Boost 1.72 (boost.org) +* Protocol Buffers 3.7.1 (https://github.com/protocolbuffers/protobuf/releases) +* CMake 3.19 or newer (cmake.org) +* Visual Studio 2019 (visualstudio.com) +* Windows SDK 8.1 (optional, if building CPU rate control for the container executor. Get this from + http://msdn.microsoft.com/en-us/windows/bg162891.aspx) +* Zlib (zlib.net, if building native code bindings for zlib) +* Git (preferably, get this from https://git-scm.com/download/win since the package also contains + Unix command-line tools that are needed during packaging). +* Python (python.org, for generation of docs using 'mvn site') * Internet connection for first build (to fetch all Maven and Hadoop dependencies) -* Unix command-line tools from GnuWin32: sh, mkdir, rm, cp, tar, gzip. These - tools must be present on your PATH. -* Python ( for generation of docs using 'mvn site') - -Unix command-line tools are also included with the Windows Git package which -can be downloaded from http://git-scm.com/downloads - -If using Visual Studio, it must be Professional level or higher. -Do not use Visual Studio Express. It does not support compiling for 64-bit, -which is problematic if running a 64-bit system. - -The Windows SDK 8.1 is available to download at: - -http://msdn.microsoft.com/en-us/windows/bg162891.aspx - -Cygwin is not required. ---------------------------------------------------------------------------------- + +Building guidelines: + +Hadoop repository provides the Dockerfile for building Hadoop on Windows 10, located at +dev-support/docker/Dockerfile_windows_10. It is highly recommended to use this and create the +Docker image for building Hadoop on Windows 10, since you don't have to install anything else +other than Docker and no additional steps are required in terms of aligning the environment with +the necessary paths etc. + +However, if you still prefer taking the route of not using Docker, this Dockerfile_windows_10 will +still be immensely useful as a raw guide for all the steps involved in creating the environment +needed to build Hadoop on Windows 10. + +Building using the Docker: +We first need to build the Docker image for building Hadoop on Windows 10. Run this command from +the root of the Hadoop repository. +> docker build -t hadoop-windows-10-builder -f .\dev-support\docker\Dockerfile_windows_10 .\dev-support\docker\ + +Start the container with the image that we just built. +> docker run --rm -it hadoop-windows-10-builder + +You can now clone the Hadoop repo inside this container and proceed with the build. + +NOTE: +While one may perceive the idea of mounting the locally cloned (on the host filesystem) Hadoop +repository into the container (using the -v option), we have seen the build to fail owing to some +files not being able to be located by Maven. Thus, we suggest cloning the Hadoop repository to a +non-mounted folder inside the container and proceed with the build. When the build is completed, +you may use the "docker cp" command to copy the built Hadoop tar.gz file from the docker container +to the host filesystem. If you still would like to mount the Hadoop codebase, a workaround would +be to copy the mounted Hadoop codebase into another folder (which doesn't point to a mount) in the +container's filesystem and use this for building. + +However, we noticed no build issues when the Maven repository from the host filesystem was mounted +into the container. One may use this to greatly reduce the build time. Assuming that the Maven +repository is located at D:\Maven\Repository in the host filesystem, one can use the following +command to mount the same onto the default Maven repository location while launching the container. +> docker run --rm -v D:\Maven\Repository:C:\Users\ContainerAdministrator\.m2\repository -it hadoop-windows-10-builder + Building: Keep the source code tree in a short path to avoid running into problems related @@ -540,6 +567,24 @@ configure the bit-ness of the build, and set several optional components. Several tests require that the user must have the Create Symbolic Links privilege. +To simplify the installation of Boost, Protocol buffers, OpenSSL and Zlib dependencies we can use +vcpkg (https://github.com/Microsoft/vcpkg.git). Upon cloning the vcpkg repo, checkout the commit +7ffa425e1db8b0c3edf9c50f2f3a0f25a324541d to get the required versions of the dependencies +mentioned above. +> git clone https://github.com/Microsoft/vcpkg.git +> cd vcpkg +> git checkout 7ffa425e1db8b0c3edf9c50f2f3a0f25a324541d +> .\bootstrap-vcpkg.bat +> .\vcpkg.exe install boost:x64-windows +> .\vcpkg.exe install protobuf:x64-windows +> .\vcpkg.exe install openssl:x64-windows +> .\vcpkg.exe install zlib:x64-windows + +Set the following environment variables - +(Assuming that vcpkg was checked out at C:\vcpkg) +> set PROTOBUF_HOME=C:\vcpkg\installed\x64-windows +> set MAVEN_OPTS=-Xmx2048M -Xss128M + All Maven goals are the same as described above with the exception that native code is built by enabling the 'native-win' Maven profile. -Pnative-win is enabled by default when building on Windows since the native components @@ -557,6 +602,24 @@ the zlib 1.2.7 source tree. http://www.zlib.net/ + +Build command: +The following command builds all the modules in the Hadoop project and generates the tar.gz file in +hadoop-dist/target upon successful build. Run these commands from an +"x64 Native Tools Command Prompt for VS 2019" which can be found under "Visual Studio 2019" in the +Windows start menu. If you're using the Docker image from Dockerfile_windows_10, you'll be +logged into "x64 Native Tools Command Prompt for VS 2019" automatically when you start the +container. + +> set classpath= +> set PROTOBUF_HOME=C:\vcpkg\installed\x64-windows +> mvn clean package -Dhttps.protocols=TLSv1.2 -DskipTests -DskipDocs -Pnative-win,dist^ + -Drequire.openssl -Drequire.test.libhadoop -Pyarn-ui -Dshell-executable=C:\Git\bin\bash.exe^ + -Dtar -Dopenssl.prefix=C:\vcpkg\installed\x64-windows^ + -Dcmake.prefix.path=C:\vcpkg\installed\x64-windows^ + -Dwindows.cmake.toolchain.file=C:\vcpkg\scripts\buildsystems\vcpkg.cmake -Dwindows.cmake.build.type=RelWithDebInfo^ + -Dwindows.build.hdfspp.dll=off -Dwindows.no.sasl=on -Duse.platformToolsetVersion=v142 + ---------------------------------------------------------------------------------- Building distributions: