1.概述
Scribe是Facebook开源的日志收集系统,在Facebook内部已经得到大量的应用。它能从各种日志源收集日志,存储到一个*存储系统上,便于进行集中统计分析处理。它为日志的”分布式收集,统一处理”提供了一个可扩展的,高容错的方案。scribe代码很简单,但是安装配置却很复杂,本文记录了作者实际的一次安装的过程,感觉真是不一般的琐碎,另外Scribe开源社区的版本已经是几年前的版本了,最新的维护信息一致没有看到,HDFS和Thrift的版本使用的都是比较旧的版本,考虑另开一个分支,升级一下Scribe,看到时候有没有时间了。
2.Scribe 安装
Scribe 依赖于多个环境,有pthyon,boost,thirft,libevent,openssl等。在编译的时候还有一些编译错误需要进行修正,安装的和准备如下。
2.1 环境准备
1.安装编译环境
#yum install gcc-c++ libevent libevent-devel automake autoconf m4 bison zlib zlib-devel bzip2 bzip2-devel flex pkgconfig python python-devel ruby ruby-devel mono-devel libxml2 libxml2-devel ant openssl-devel
2.安装boost环境:最低版本1.36
#wget http://nchc.dl.sourceforge.net/project/boost/boost/1.45.0/boost_1_45_0.tar.gz
#tar zxvf boost_1_45_0.tar.gz
#cd boost_1_45_0
#./bootstrap.sh
#./bjam
#./bjam --s HAVE_ICU=1 --prefix=/usr/local/boost --includedir=/usr/local/boost/include --libdir=/usr/local/boost/lib
#./bjam install –prefix=/usr/local/boost
#echo "/usr/local/boost/lib" >> /etc/ld.so.conf
#echo /usr/local/boost/include >> /etc/ld.so.conf
#ldconfig
3. 安装thrift环境
#wget https://dist.apache.org/repos/dist/release/thrift/0.9.0/thrift-0.9.0.tar.gz
#tar xzvf thrift-0.9.0.tar.gz
#./configure --prefix=/usr/local/thrift --with-boost=/usr/local/boost --with-java=no --with-erlang=no --with-perl=no --with-php=no --with-ruby=no
备注:如果提示找不到libcrpty增加openssl路径
./configure --prefix=/usr/local/thrift --with-boost=/usr/local/boost --with-java=no --with-erlang=no --with-perl=no --with-php=no --with- ruby=no CPPFLAGS="-I/usr/local/openssl/include" LDFLAGS="-ldl -L/usr/local/openssl/lib"
修改文件:
Thrift/lib/ src/thrift/Thrift.cpp 增加头文件:#include<stdlib.h>
同时修改config.h文件,注释掉:
/* Define to rpl_malloc if the replacement function should be used. */
//modby zhangzl++:
//#define malloc rpl_malloc
//--
…
/* Define to rpl_realloc if the replacement function should be used. */
//modby zhangzl++:
//#define realloc rpl_realloc
//--
#make
#make install
# echo "/usr/local/thrift/lib" >> /etc/ld.so.conf
#ldconfig
#cd contrib.
4.安装fb303
#cd fb303
#./bootstrap.sh
#./configure --prefix=/usr/local/thrift/fb303 --with-boost=/usr/local/boost/ --with-thriftpath=/usr/local/thrift
#make
//注意修改观察生成的Makefile关于头文件和库文件路径的信息是否正确,如果正确则可以直接编译。
#make install
4.安装scribe
在https://github.com/facebook/scribe下载zip包。
#unzip scribe-master.zip
#cd scribe-master
# export BOOST_ROOT=/usr/local/boost
#export LD_LIBRARY_PATH=/usr/local/thrift/lib:/usr/lib:/usr/local/lib:/usr/local/boost/lib:/usr/local/thrift/fb303/lib
#./bootstrap.sh --with-boost=/usr/local/boost --with-boost-filesystem=boost_filesystem
遇到问题:
checking whether the Boost::System library is available… yes
checking whether the Boost::Filesystem library is available… yes
configure: error: Could not link against !
解决办法,在configure 后加一个参数 --with-boost-filesystem=boost_filesystem
#./configure --prefix=/usr/local/scribe --with-boost=/usr/local/boost --with-thriftpath=/usr/local/thrift --with-fb303path=/usr/local/thrift/fb303 --with-boost-filesystem=boost_filesystem
#make
遇到问题:
- thrift/server/TNonblockingServer.h:40:19: error: event.h: No such file or directory
- 安装Libevent
#tar zxvf libevent-2.0.18-stable.tar.gz
#cd libevent-2.0.18-stable
#./configure –prefix=/usr/local/libevent
#make
#make install
3.安装libevent-devel
yum install libevent-devel
error: there are no arguments to 'htons' that depend on a template parameter, so a declaration of 'htons' must be available
需要自己修改一下文件:
vim /usr/local/thrift/include/thrift/protocol/TBinaryProtocol.tcc
在首行添加头文件引用: #include <arpa/inet.h>
#echo /usr/local/libevent/include>> /etc/ld.so.conf
#ldconfig
#make install
2.2 测试
#cp example/example1.conf /tmp
#/usr/local/scribe/bin/scribed /tmp/example1.conf
进行测试。
如果失败,根据提示信息,进行相应修改。
我启动时出现下面的错误:
./scribed: error while loading shared libraries: libevent-2.0.so.5: cannot open shared object file: No such file or directory
解决办法是find这个库文件libevent-2.0.so.5,拷贝到/usr/lib64目录下。
启动成功后,出现下面的界面:
3.Scribe配置
Scribe 的配置解析见http://my.oschina.net/guol/blog/110258
测试通过的server端和client端的配置文件如下:
- Server.conf
port=1463
max_msg_per_second=100000000
max_queue_size=10000000
check_interval=1
new_thread_per_category=true
<store>
category=squid
type=file
file_path=/tmp/test/logcollect/
base_filename=21vianet__squid_access_all_
rotate_period=daily
rotate_hour=0
rotate_minute=0
max_size=1610612736
add_newlines=0
</store>
- Client.conf
port=1464
max_msg_per_second=2000000
max_queue_size=1000000
check_interval=1
new_thread_per_category=true
<store>
category=default
type=buffer
target_write_size=20480
max_write_interval=1
buffer_send_rate=2
retry_interval=30
retry_interval_range=10
<primary>
type=network
#remote_host=218.93.205.106
remote_host=192.168.11.95
remote_port=1463
</primary>
<secondary>
type=file
fs_type=std
write_meta=no
file_path=/tmp/test/scribed
base_filename=accesslog
max_size=100000000
add_newlines=0
</secondary>
</store>
注意:
在server端接收的日志,将以文件的方式来记录log,这个文件不能删除,它是个软连接。需要先copy过去,再调用touch命令将文件清空。