Files

284 lines
11 KiB
Tcl

# -*- coding: utf-8; mode: tcl; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- vim:fenc=utf-8:ft=tcl:et:sw=4:ts=4:sts=4
PortSystem 1.0
name hadoop
version 1.2.1
revision 1
categories java devel science
maintainers nomaintainer
description Open-source software for reliable, scalable, distributed computing
long_description Hadoop is a distributed computing platform written in Java. \
It incorporates features similar to those of the Google File System \
and of MapReduce.
homepage https://hadoop.apache.org/
platforms darwin
license Apache-2
master_sites apache:hadoop/common/${distname}
checksums rmd160 6330ded6043a1c8dd8597910e77f3dedf249a807 \
sha256 94a1181771f173bdb55c8f901722825866396091f0516bdd12b34dc3de1706a1
patchfiles patch-conf-hadoop-env.sh.diff \
patch-src-native.diff \
patch-src-c++.diff
depends_build bin:ant:apache-ant \
port:autoconf port:automake port:libtool
depends_lib port:zlib \
port:snappy
set java_home /System/Library/Frameworks/JavaVM.framework/Versions/1.6/Home
# This port works on Mac OS X 10.6 or later, because 'sudo option -E'
# is not available on Mac OS X 10.5 or earlier. See #34665.
if {${os.platform} eq "darwin" && ${os.major} <= 9} {
known_fail yes
pre-fetch {
ui_error "This port works on Mac OS X 10.6 (Snow Leopard) or later"
return -code error "Mac OS X 10.6 (Snow Leopard) or later is required"
}
}
pre-fetch {
if {![file exists ${java_home}]} {
ui_error "Java 1.6 is required, but not located at ${java_home}"
return -code error "Java 1.6 missing"
}
}
# Fix a file permission to build c++-libhdfs.
post-extract {
file attributes ${worksrcpath}/src/c++/libhdfs/install-sh -permissions 0755
}
variant universal {}
use_configure no
set java_include -I/System/Library/Frameworks/JavaVM.framework/Headers
set cflags "${configure.cflags} [get_canonical_archflags]"
set cxxflags "${configure.cxxflags} [get_canonical_archflags cxx]"
set ldflags "${configure.ldflags} [get_canonical_archflags] -framework JavaVM"
set cppflags "${configure.cppflags} ${java_include}"
# Set configure args to build native and c++-libhdfs.
post-patch {
set libs "-ldl -lz -lsnappy"
reinplace "s|@cc@|${configure.cc}|g" ${worksrcpath}/build.xml
reinplace "s|@cflags@|${cflags}|g" ${worksrcpath}/build.xml
reinplace "s|@cxx@|${configure.cxx}|g" ${worksrcpath}/build.xml
reinplace "s|@cxxflags@|${cxxflags}|g" ${worksrcpath}/build.xml
reinplace "s|@ldflags@|${ldflags}|g" ${worksrcpath}/build.xml
reinplace "s|@libs@|${libs}|g" ${worksrcpath}/build.xml
reinplace "s|@cppflags@|${cppflags}|g" ${worksrcpath}/build.xml
reinplace "s|@args@|--prefix=${prefix}|g" ${worksrcpath}/build.xml
}
# Build native and c++-libhdfs.
build.cmd ant
build.args -Dcompile.native=true \
-Dsnappy.prefix=${prefix} \
-Dcompile.c++=true \
-Dlibhdfs=true
build.target compile-native compile-c++-libhdfs
build.env JAVA_HOME=${java_home}
# Fix install_name of dylib.
pre-destroot {
foreach file [glob ${worksrcpath}/build/native/**/lib/*.dylib \
${worksrcpath}/build/c++/**/lib/*.dylib] {
if {[file isfile ${file}]} {
set libname [file tail ${file}]
system "install_name_tool -id ${prefix}/lib/${libname} ${file}"
}
}
}
# Hadoop home and conf directories.
set hadoop_basedir ${prefix}/share/java
set hadoop_home ${hadoop_basedir}/${distname}
set hadoop_conf_dir ${hadoop_home}/conf
# Document directory.
set hadoop_doc_dir ${prefix}/share/doc/${distname}
# Working directories.
set hadoop_var_dir ${prefix}/var/${name}
set hadoop_log_dir ${hadoop_var_dir}/log
set hadoop_pid_dir ${hadoop_var_dir}/run
set hadoopuser hadoop
add_users ${hadoopuser} \
group=${hadoopuser} \
realname=Hadoop\ Server \
home=${hadoop_var_dir} \
shell=/bin/bash
variant fusedfs description {Add Fuse-DFS} {
depends_lib-append port:osxfuse
patchfiles-append patch-src-contrib-fusedfs.diff
# libhdfs.dylib must be built before configuring fuse-dfs.
post-build {
set libs "-lfuse -lhdfs"
# "$@" is replaced with "${hadoop.root}/build/c++/${build.platform}/lib".
# See files/patch-src-contrib-fusedfs.diff.
set args "--prefix=${prefix} \
CC=${configure.cc} \
CFLAGS=\"${cflags}\" \
LDFLAGS=\"${ldflags} -L$@\" \
LIBS=\"${libs}\" \
CPPFLAGS=\"${cppflags}\""
set sh ${worksrcpath}/src/contrib/fuse-dfs/bootstrap.sh
reinplace "s|\./configure|\./configure ${args}|" ${sh}
# Build fusedfs.
system -W ${worksrcpath} "ant compile-contrib -Dlibhdfs=1 -Dfusedfs=1"
}
post-destroot {
xinstall -m 755 \
${worksrcpath}/build/contrib/fuse-dfs/fuse_dfs \
${destroot}${prefix}/bin
# Fix install_name in fuse_dfs.
set bin ${destroot}${prefix}/bin/fuse_dfs
regexp {(\S+\/libhdfs\S+dylib)} [exec otool -L ${bin}] path
system "install_name_tool -change ${path} ${prefix}/lib/libhdfs.dylib ${bin}"
# Install fuse_dfs_wrapper.sh.
xinstall -m 755 ${filespath}/fuse_dfs_wrapper.sh ${destroot}${hadoop_home}/bin
set sh ${destroot}${hadoop_home}/bin/fuse_dfs_wrapper.sh
reinplace "s|@hadoop_home@|${hadoop_home}|" ${sh}
reinplace "s|@java_home@|${java_home}|" ${sh}
reinplace "s|@prefix@|${prefix}|" ${sh}
}
}
destroot {
# Copy the distribution to Hadoop home directory.
xinstall -m 755 -d ${destroot}${hadoop_home}
foreach dir {bin conf contrib lib libexec sbin share webapps} {
copy ${worksrcpath}/${dir} ${destroot}${hadoop_home}
}
foreach file [glob ${worksrcpath}/hadoop-*] {
xinstall -m 644 ${file} ${destroot}${hadoop_home}
}
delete ${destroot}${hadoop_home}/lib/native
delete ${destroot}${hadoop_home}/libexec/jsvc.amd64
# Patch for Mahout 0.4 to suppress warnings.
set webinf ${destroot}${hadoop_home}/webapps/secondary/WEB-INF
xinstall -m 755 -d ${webinf}
destroot.keepdirs-append ${webinf}
# Install native and c++ libraries.
foreach file [glob ${worksrcpath}/build/native/**/lib/* \
${worksrcpath}/build/c++/**/lib/*] {
xinstall -m 644 ${file} ${destroot}${prefix}/lib
}
# Copy extra documents.
xinstall -m 755 -d ${destroot}${hadoop_doc_dir}
copy ${worksrcpath}/docs ${destroot}${hadoop_doc_dir}
foreach file [glob ${worksrcpath}/*.txt] {
xinstall -m 644 ${file} ${destroot}${hadoop_doc_dir}
}
# Install an extra script for this port.
set hadoop_bin ${destroot}${prefix}/bin/hadoop-bin
xinstall -m 755 ${filespath}/hadoop-bin ${hadoop_bin}
reinplace "s|@hadoop_home@|${hadoop_home}|g" ${hadoop_bin}
reinplace "s|@java_home@|${java_home}|g" ${hadoop_bin}
reinplace "s|@hadoopuser@|${hadoopuser}|g" ${hadoop_bin}
# Setup 'hadoop-env.sh' in conf.
set env_sh ${destroot}${hadoop_conf_dir}/hadoop-env.sh
reinplace "s|@java_home@|${java_home}|g" ${env_sh}
reinplace "s|@hadoop_log_dir@|${hadoop_log_dir}|g" ${env_sh}
reinplace "s|@hadoop_pid_dir@|${hadoop_pid_dir}|g" ${env_sh}
# Create working directories.
xinstall -m 755 -o ${hadoopuser} -g ${hadoopuser} -d \
${destroot}${hadoop_var_dir} \
${destroot}${hadoop_log_dir} \
${destroot}${hadoop_pid_dir}
destroot.keepdirs-append \
${destroot}${hadoop_var_dir} \
${destroot}${hadoop_log_dir} \
${destroot}${hadoop_pid_dir}
}
post-deactivate {
ui_msg "********************************************************"
ui_msg "* To revert the system after uninstalling the port:"
ui_msg "* 1) Delete Hadoop working directory:"
ui_msg "* $ sudo rm -rf ${hadoop_var_dir}"
ui_msg "* 2) Delete Hadoop user and group:"
ui_msg "* $ sudo dscl . -delete /Users/${hadoopuser}"
ui_msg "* $ sudo dscl . -delete /Groups/${hadoopuser}"
ui_msg "********************************************************"
}
default_variants +pseudo
set hadoop_tmp_dir ${hadoop_var_dir}/cache
variant pseudo description {Run on a single-node in a pseudo-distributed mode} {
patchfiles-append patch-pseudo.diff
post-destroot {
# Set conf directory for a pseudo-distributed mode.
copy ${destroot}${hadoop_conf_dir} ${destroot}${hadoop_conf_dir}.pseudo
move ${destroot}${hadoop_conf_dir} ${destroot}${hadoop_conf_dir}.local
ln -s conf.pseudo ${destroot}${hadoop_conf_dir}
# Set the maximum number of tasks based on the number of the CPUs (cores).
regexp {\d+} [exec sysctl hw.ncpu] ncpu
set tasks_max [expr $ncpu + 2]
# Setup configuration files.
reinplace "s|@tasks_max@|${tasks_max}|g" \
${destroot}${hadoop_conf_dir}.pseudo/mapred-site.xml
reinplace "s|@hadoop_tmp_dir@|${hadoop_tmp_dir}|" \
${destroot}${hadoop_conf_dir}.pseudo/core-site.xml
# Create a temporary directory.
xinstall -m 755 -o ${hadoopuser} -g ${hadoopuser} -d \
${destroot}${hadoop_tmp_dir}
destroot.keepdirs-append \
${destroot}${hadoop_tmp_dir}
}
post-activate {
# Setup passphraseless ssh.
set ssh_dir ${hadoop_var_dir}/.ssh
if {![file exists ${ssh_dir}]} {
xinstall -m 700 -o ${hadoopuser} -g ${hadoopuser} -d ${ssh_dir}
system "sudo -u ${hadoopuser} ssh-keygen -t rsa -P '' -f ${ssh_dir}/id_rsa"
xinstall -m 644 -o ${hadoopuser} -g ${hadoopuser} \
${ssh_dir}/id_rsa.pub \
${ssh_dir}/authorized_keys
}
ui_msg "********************************************************"
ui_msg "* To run on a single-node in a pseudo-distributed mode:"
ui_msg "* 1) Turn on Remote Login (sshd):"
ui_msg "* check 'System Preferences > Sharing > Remote Login'"
ui_msg "* 2) Format a new distributed-filesystem:"
ui_msg "* $ hadoop-bin hadoop namenode -format"
ui_msg "* 3) Start the hadoop daemons:"
ui_msg "* $ hadoop-bin start-all.sh"
ui_msg "* 4) Perform operations you like. To see examples:"
ui_msg "* $ open file://${hadoop_home}/docs/single_node_setup.html"
ui_msg "* 5) When you're done, stop the daemons with:"
ui_msg "* $ hadoop-bin stop-all.sh"
ui_msg "********************************************************"
}
}
livecheck.type regex
livecheck.url https://www.apache.org/dist/hadoop/common/stable1/
livecheck.regex ${name}-(\[0-9.\]+).tar