您正在查看: 2018年4月

R语言版本LightGBM在ubuntu16.04上的安装

Overview

首先非常感谢谢若鹏同学给的LightGBM安装教程和调优等脚本。下午在自己的6G内存ubuntu16.04系统上安装,在make -j这一步编译c++boosting库时总是退出,提示虚拟内存不足,看来是电脑配置太低了。只能在Bastion3服务器上面测试了。

1.LightGBM的安装

首先安装git

sudo apt-get install git

github上面clone``LightGBM的源码:

git clone --recursive https://github.com/Microsoft/LightGBM

然后执行以下几步:

cd LightGBM  
mkdir build  
cd build

下面先安装一下两个依赖包:

sudo apt-get install cmake
sudo apt-get install make

继续执行:

cmake ..
make -j

安装完成即可。

这一步对电脑内存要求比较高,我6G的内存都不够,坑。

2.R包的安装

大约需要以下几步安装:

install.packages("readr")
install.packages("GA")
install.packages("dplyr")
install.packages("parallelMap")
install.packages("jsonlite")
install.packages("mlr")

提示一下:在安装mlr包之前必须用以下命令安装好两个系统依赖包:

sudo apt-get install libxml2
sudo apt-get install libxml2-dev

否则会提示:

ERROR: dependency ‘XML’ is not available for package ‘mlr’

继续安装LightGBM需要的R包:

cd LightGBM/R-package
Rscript build_package.R
sudo R CMD INSTALL lightgbm_2.1.0.tar.gz --no-multiarch

至此,R语言版本的LightGBM算是安装好了。

本文参考了以下文章,感谢!
案例 | lightgbm算法优化-不平衡二分类问题(附代码)
Ubuntu 14.04下libxml2的安装和使用

FastDFS分布式文件系统在服务器集群上的安装部署

Overview

到目前为止,我们手里已经有了10台服务器了。之前一直说要把这些服务器搭建一个分布式文件系统,现在条件终于成熟了。这些服务器预装的系统大多数是Ubuntu16.04LTS发行版的Linux,少部分是Ubuntu14.04LTS。这次我们选择5台服务器,1feagen(118.138.241.39)服务器作为tracker4台服务器(Bastion3(118.138.240.146),Bastion4(118.138.233.74),Bastion6(118.138.233.26),POSSUM(118.138.233.27))作为storage。这里会记录下FastDFS分布式文件系统在Ubuntu16.04上的安装部署过程的细节。

5台服务器都要安装,只是配置文件稍有不同。

1.必备软件

  • FastDFS
    这个去GitHub上面下载源码:FastDFS
  • libfastcommon
    这个仍然要去GitHub上面下载源码:libfastcommon

这两个Zip包用wget或者在本地下载上传至服务器均可。

2.libfastcommon的安装

首先解压libfastcommon-master.zip包:

unzip libfastcommon-master.zip

进入这个文件夹:

cd libfastcommon-master/

依次输入如下命令:

sudo ./make.sh
sudo ./make.sh install

输入第二条命令之后,会有如下显示:

mkdir -p /usr/lib64
mkdir -p /usr/lib
install -m 755 libfastcommon.so /usr/lib64
install -m 755 libfastcommon.so /usr/lib
mkdir -p /usr/include/fastcommon
install -m 644 common_define.h hash.h chain.h logger.h base64.h shared_func.h pthread_func.h ini_file_reader.h _os_define.h sockopt.h sched_thread.h http_func.h md5.h local_ip_func.h avl_tree.h ioevent.h ioevent_loop.h fast_task_queue.h fast_timer.h process_ctrl.h fast_mblock.h connection_pool.h fast_mpool.h fast_allocator.h fast_buffer.h skiplist.h multi_skiplist.h flat_skiplist.h skiplist_common.h system_info.h fast_blocked_queue.h php7_ext_wrapper.h id_generator.h char_converter.h char_convert_loader.h /usr/include/fastcommon

可以看到libfastcommon.so安装到了/usr/lib64.
下面,我们需要为两个文件创建软链接,指向FastDFS主程序的lib目录:

sudo ln -s /usr/lib64/libfastcommon.so /usr/local/lib/libfastcommon.so
sudo ln -s /usr/lib64/libfastcommon.so /usr/lib/libfastcommon.so
sudo ln -s /usr/lib64/libfdfsclient.so /usr/local/lib/libfdfsclient.so
sudo ln -s /usr/lib64/libfdfsclient.so /usr/lib/libfdfsclient.so

3.FastDFS的安装

解压fastdfs-master.zip文件:

unzip fastdfs-master.zip

进入FastDFS源码根目录:

cd fastdfs-master

依次输入下面两条命令:

sudo ./make.sh
sudo ./make.sh install

安装完成之后,所有的可执行文件都被放在了/usr/bin/目录下面,可以用下面的命令查看:

ls /usr/bin/fdfs*

显示共有14个文件:

/usr/bin/fdfs_appender_test   
/usr/bin/fdfs_download_file  
/usr/bin/fdfs_test1
/usr/bin/fdfs_appender_test1  
/usr/bin/fdfs_file_info      
/usr/bin/fdfs_trackerd
/usr/bin/fdfs_append_file     
/usr/bin/fdfs_monitor        
/usr/bin/fdfs_upload_appender
/usr/bin/fdfs_crc32           
/usr/bin/fdfs_storaged       
/usr/bin/fdfs_upload_file
/usr/bin/fdfs_delete_file     
/usr/bin/fdfs_test

4. 配置FastDFS

配置文件在目录/etc/fdfs下,查看一下:

ls /etc/fdfs

显示如下:

client.conf.sample  storage.conf.sample  storage_ids.conf.sample  tracker.conf.sample

4.1 创建跟踪器和存储节点的配置文件

cd /etc/fdfs
sudo cp tracker.conf.sample tracker.conf
sudo cp storage.conf.sample storage.conf

4.2 修改tracker配置文件

feagen(118.138.241.39)服务器中修改tracker配置文件如下:

sudo vim /etc/fdfs/tracker.conf

可以看到如下显示:

# the base path to store data and log files
base_path=/home/yuqing/fastdfs

这行改成如下内容:

base_path=/feagen/fastdfs/tracker

这种配置文件,注意等号两边不能有空格。且上面的目录必须是真实存在的。
启动tracker

sudo fdfs_trackerd /etc/fdfs/tracker.conf start

查看监听端口:

sudo netstat -unltp|grep fdfs

如果显示如下,则证明tracker启动成功:

tcp        0      0 0.0.0.0:22122           0.0.0.0:*               LISTEN      6951/fdfs_trackerd

4.3 修改storage配置文件

Bastion3(118.138.240.146)服务器中,修改storage配置文件如下:

sudo vim /etc/fdfs/storage.conf

可以看到如下显示:

group_name=group1
base_path=/home/yuqing/fastdfs
store_path0=/home/yuqing/fastdfs
tracker_server=192.168.209.121:22122

修改为:

group_name=group1
base_path=/bastion3_cache/fastdfs/storage
store_path0=/bastion3_cache/fastdfs/storage
tracker_server=118.138.241.39:22122

Bastion4(118.138.233.74)服务器中,修改storage配置文件如下:

sudo vim /etc/fdfs/storage.conf

可以看到如下显示:

group_name=group1
base_path=/home/yuqing/fastdfs
store_path0=/home/yuqing/fastdfs
tracker_server=192.168.209.121:22122

修改为:

group_name=group1
base_path=/bastion4_cache/fastdfs/storage
store_path0=/bastion4_cache/fastdfs/storage
tracker_server=118.138.241.39:22122

Bastion6(118.138.233.26)服务器中,修改storage配置文件如下:

sudo vim /etc/fdfs/storage.conf

可以看到如下显示:

group_name=group1
base_path=/home/yuqing/fastdfs
store_path0=/home/yuqing/fastdfs
tracker_server=192.168.209.121:22122

修改为:

group_name=group1
base_path=/bastion6_cache/fastdfs/storage
store_path0=/bastion6_cache/fastdfs/storage
tracker_server=118.138.241.39:22122

POSSUM(118.138.233.27)服务器中,修改storage配置文件如下:

sudo vim /etc/fdfs/storage.conf

可以看到如下显示:

group_name=group1
base_path=/home/yuqing/fastdfs
store_path0=/home/yuqing/fastdfs
tracker_server=192.168.209.121:22122

修改为:

group_name=group1
base_path=/possum/fastdfs/storage
store_path0=/possum/fastdfs/storage
tracker_server=118.138.241.39:22122

启动storage服务:

sudo fdfs_storaged /etc/fdfs/storage.conf start

显示如下,即代表成功启动:

process fdfs_storaged already running, pid: 28250

查看监听端口:

sudo netstat -unltp|grep fdfs

如果显示如下,则证明storage启动成功:

tcp        0      0 0.0.0.0:23000           0.0.0.0:*               LISTEN      28250/fdfs_storaged

如果没有显示,则有可能是storage启动失败。

那么我们以Bastion6服务器为例,查看启动日志:

tail /bastion6_cache/fastdfs/storage/logs/storaged.log

如果显示如下,代表启动失败:

[2018-05-13 19:56:31] ERROR - file: storage_ip_changed_dealer.c, line: 186, connect to tracker server 118.138.2
40.146:22122 fail, errno: 110, error info: Connection timed out

连接超时;
如果显示如下,代表启动成功:

[2018-05-14 12:59:40] INFO - file: tracker_client_thread.c, line: 310, successfully connect to tracker server 118.138.240.146:22122, as a tracker client, my ip is 118.138.233.26
[2018-05-14 13:00:10] INFO - file: tracker_client_thread.c, line: 1263, tracker server 118.138.240.146:22122, set tracker leader: 118.138.240.146:22122
[2018-05-14 13:03:06] INFO - file: storage_sync.c, line: 2733, successfully connect to storage server 118.138.233.74:23000, continuous fail count: 16
[2018-05-14 13:03:41] INFO - file: storage_sync.c, line: 2733, successfully connect to storage server 118.138.233.74:23000

连接成功。

超时解读:
之前跟Chris讨论的时候,刚开始以为这几天服务器都是Monash云服务器中心的,属于内网,所以相互之间的端口是不用开的,实验之后才知道不开端口是不行的。后来联系Jerico开开3个端口之后,才连接成功。这3个端口分别是22122888823000

  • 22122:代表tracker服务端口;
  • 8888:代表HTTP协议端口,网页上传下载文件需要这个端口;
  • 23000:代表storage服务端口。

至此,storage存储节点安装成功。

4.4 查看所有存储节点信息

所有存储节点都启动之后,在任意一台storage上面用下面命令,查看集群状态信息:

sudo /usr/bin/fdfs_monitor /etc/fdfs/storage.conf

显示如下:

[2018-05-24 01:27:18] DEBUG - base_path=/bastion6_cache/fastdfs/storage, connect_timeout=10, network_timeout=60, tracker_server_count=1, anti_steal_token=0, anti_steal_secret_key length=0, use_connection_pool=0, g_connection_pool_max_idle_time=3600s, use_storage_id=0, storage server id count: 0

server_count=1, server_index=0

tracker server is 118.138.241.39:22122

group count: 1

Group 1:
group name = group1
disk total space = 100665 MB
disk free space = 72339 MB
trunk free space = 0 MB
storage server count = 4
active server count = 4
storage server port = 23000
storage HTTP port = 8888
store path count = 1
subdir count per path = 256
current write server index = 0
current trunk file id = 0

    Storage 1:
        id = 118.138.233.26
        ip_addr = 118.138.233.26 (vm-118-138-233-26.erc.monash.edu.au)  ACTIVE
        http domain = 
        version = 5.12
        join time = 2018-05-13 19:55:46
        up time = 2018-05-20 00:51:17
        total storage = 483679 MB
        free storage = 452554 MB
        upload priority = 10
        store_path_count = 1
        subdir_count_per_path = 256
        storage_port = 23000
        storage_http_port = 8888
        current_write_path = 0
        source storage id = 
        if_trunk_server = 0
        connection.alloc_count = 256
        connection.current_count = 3
        connection.max_count = 3
        total_upload_count = 1
        success_upload_count = 1
        total_append_count = 0
        success_append_count = 0
        total_modify_count = 0
        success_modify_count = 0
        total_truncate_count = 0
        success_truncate_count = 0
        total_set_meta_count = 0
        success_set_meta_count = 0
        total_delete_count = 0
        success_delete_count = 0
        total_download_count = 0
        success_download_count = 0
        total_get_meta_count = 0
        success_get_meta_count = 0
        total_create_link_count = 0
        success_create_link_count = 0
        total_delete_link_count = 0
        success_delete_link_count = 0
        total_upload_bytes = 87026
        success_upload_bytes = 87026
        total_append_bytes = 0
        success_append_bytes = 0
        total_modify_bytes = 0
        success_modify_bytes = 0
        stotal_download_bytes = 0
        success_download_bytes = 0
        total_sync_in_bytes = 11288
        success_sync_in_bytes = 11288
        total_sync_out_bytes = 0
        success_sync_out_bytes = 0
        total_file_open_count = 5
        success_file_open_count = 5
        total_file_read_count = 0
        success_file_read_count = 0
        total_file_write_count = 5
        success_file_write_count = 5
        last_heart_beat_time = 2018-05-24 01:26:54
        last_source_update = 2018-05-17 01:27:29
        last_sync_update = 2018-05-17 01:06:46
        last_synced_timestamp = 1970-01-01 10:00:00 (never synced)
    Storage 2:
        id = 118.138.233.27
        ip_addr = 118.138.233.27 (vm-118-138-233-27.erc.monash.edu.au)  ACTIVE
        http domain = 
        version = 5.12
        join time = 2018-05-17 00:39:17
        up time = 2018-05-20 00:52:30
        total storage = 2015737 MB
        free storage = 1689979 MB
        upload priority = 10
        store_path_count = 1
        subdir_count_per_path = 256
        storage_port = 23000
        storage_http_port = 8888
        current_write_path = 0
        source storage id = 
        if_trunk_server = 0
        connection.alloc_count = 256
        connection.current_count = 3
        connection.max_count = 3
        total_upload_count = 0
        success_upload_count = 0
        total_append_count = 0
        success_append_count = 0
        total_modify_count = 0
        success_modify_count = 0
        total_truncate_count = 0
        success_truncate_count = 0
        total_set_meta_count = 0
        success_set_meta_count = 0
        total_delete_count = 0
        success_delete_count = 0
        total_download_count = 0
        success_download_count = 0
        total_get_meta_count = 0
        success_get_meta_count = 0
        total_create_link_count = 0
        success_create_link_count = 0
        total_delete_link_count = 0
        success_delete_link_count = 0
        total_upload_bytes = 0
        success_upload_bytes = 0
        total_append_bytes = 0
        success_append_bytes = 0
        total_modify_bytes = 0
        success_modify_bytes = 0
        stotal_download_bytes = 0
        success_download_bytes = 0
        total_sync_in_bytes = 98314
        success_sync_in_bytes = 98314
        total_sync_out_bytes = 0
        success_sync_out_bytes = 0
        total_file_open_count = 5
        success_file_open_count = 5
        total_file_read_count = 0
        success_file_read_count = 0
        total_file_write_count = 5
        success_file_write_count = 5
        last_heart_beat_time = 2018-05-24 01:27:08
        last_source_update = 1970-01-01 10:00:00
        last_sync_update = 2018-05-21 16:32:36
        last_synced_timestamp = 2018-05-17 01:06:38 (20m:51s delay)
    Storage 3:
        id = 118.138.233.74
        ip_addr = 118.138.233.74 (vm-118-138-233-74.erc.monash.edu.au)  ACTIVE
        http domain = 
        version = 5.12
        join time = 2018-05-13 20:03:06
        up time = 2018-05-20 00:53:40
        total storage = 100665 MB
        free storage = 72339 MB
        upload priority = 10
        store_path_count = 1
        subdir_count_per_path = 256
        storage_port = 23000
        storage_http_port = 8888
        current_write_path = 0
        source storage id = 
        if_trunk_server = 0
        connection.alloc_count = 256
        connection.current_count = 3
        connection.max_count = 3
        total_upload_count = 2
        success_upload_count = 2
        total_append_count = 0
        success_append_count = 0
        total_modify_count = 0
        success_modify_count = 0
        total_truncate_count = 0
        success_truncate_count = 0
        total_set_meta_count = 2
        success_set_meta_count = 2
        total_delete_count = 0
        success_delete_count = 0
        total_download_count = 0
        success_download_count = 0
        total_get_meta_count = 0
        success_get_meta_count = 0
        total_create_link_count = 0
        success_create_link_count = 0
        total_delete_link_count = 0
        success_delete_link_count = 0
        total_upload_bytes = 11190
        success_upload_bytes = 11190
        total_append_bytes = 0
        success_append_bytes = 0
        total_modify_bytes = 0
        success_modify_bytes = 0
        stotal_download_bytes = 0
        success_download_bytes = 0
        total_sync_in_bytes = 174142
        success_sync_in_bytes = 87026
        total_sync_out_bytes = 0
        success_sync_out_bytes = 0
        total_file_open_count = 3
        success_file_open_count = 3
        total_file_read_count = 0
        success_file_read_count = 0
        total_file_write_count = 3
        success_file_write_count = 3
        last_heart_beat_time = 2018-05-24 01:27:13
        last_source_update = 2018-05-17 01:06:38
        last_sync_update = 2018-05-21 16:32:54
        last_synced_timestamp = 2018-05-17 01:27:30 (-1s delay)
    Storage 4:
        id = 118.138.240.146
        ip_addr = 118.138.240.146 (vm-118-138-240-146.erc.monash.edu.au)  ACTIVE
        http domain = 
        version = 5.12
        join time = 2018-05-20 00:53:13
        up time = 2018-05-20 00:53:13
        total storage = 483679 MB
        free storage = 449872 MB
        upload priority = 10
        store_path_count = 1
        subdir_count_per_path = 256
        storage_port = 23000
        storage_http_port = 8888
        current_write_path = 0
        source storage id = 118.138.233.26
        if_trunk_server = 0
        connection.alloc_count = 256
        connection.current_count = 3
        connection.max_count = 3
        total_upload_count = 0
        success_upload_count = 0
        total_append_count = 0
        success_append_count = 0
        total_modify_count = 0
        success_modify_count = 0
        total_truncate_count = 0
        success_truncate_count = 0
        total_set_meta_count = 0
        success_set_meta_count = 0
        total_delete_count = 0
        success_delete_count = 0
        total_download_count = 0
        success_download_count = 0
        total_get_meta_count = 0
        success_get_meta_count = 0
        total_create_link_count = 0
        success_create_link_count = 0
        total_delete_link_count = 0
        success_delete_link_count = 0
        total_upload_bytes = 0
        success_upload_bytes = 0
        total_append_bytes = 0
        success_append_bytes = 0
        total_modify_bytes = 0
        success_modify_bytes = 0
        stotal_download_bytes = 0
        success_download_bytes = 0
        total_sync_in_bytes = 98314
        success_sync_in_bytes = 98314
        total_sync_out_bytes = 0
        success_sync_out_bytes = 0
        total_file_open_count = 5
        success_file_open_count = 5
        total_file_read_count = 0
        success_file_read_count = 0
        total_file_write_count = 5
        success_file_write_count = 5
        last_heart_beat_time = 2018-05-24 01:27:03
        last_source_update = 1970-01-01 10:00:00
        last_sync_update = 2018-05-21 16:32:54
        last_synced_timestamp = 2018-05-17 01:27:30 (-1s delay)

可以看到,4storage状态都是ACTIVE,表示都启动成功了。

5. 测试上传文件

需要注意一点,我们之后的特征提取程序会在feagen服务器上,所以,我们暂时将client安装在这台服务器上。测试一下在本台服务器上上传文件,能否在所有storage服务器上面查看得到此文件的备份。

5.1 修改tracker服务器上的client配置文件

先后输入以下命令:

cd /etc/fdfs
sudo cp client.conf.sample client.conf
sudo vim client.conf

base_pathtracker_server两处配置修改为:

base_path=/feagen/fastdfs/client
tracker_server=118.138.241.39:22122

shell当中输入下面的命令:

sudo /usr/bin/fdfs_upload_file /etc/fdfs/client.conf fastdfs-master.zip

返回如下,即可表明上传成功:

group1/M00/00/00/dorpG1td37yAFSYPAAZ-Sk23ivY904.zip

这个返回值,在perl当中,以如下方式获取:

my $file_name = qx(sudo /usr/bin/fdfs_upload_file /etc/fdfs/client.conf fastdfs-master.zip);

到任意一个storage服务器当中查看,以Bastion3为例,至此处:

/bastion3_cache/fastdfs/storage/data/00/00

发现存在如下文件:

dorpG1td37yAFSYPAAZ-Sk23ivY904.zip

即表明上传成功。

6. 测试下载文件

当前目录下,运行下面的程序:

sudo /usr/bin/fdfs_download_file /etc/fdfs/client.conf group1/M00/00/00/dorpG1td37yAFSYPAAZ-Sk23ivY904.zip

即可在当前目录下面找到dorpG1td37yAFSYPAAZ-Sk23ivY904.zip,表明下载成功。

至此我们的分布式文件服务器算是搭建完成了。

我们参考了这些文章,表示感谢!
FastDFS分布式文件系统集群安装与配置
FastDFS--原理篇
分布式文件系统FastDFS原理介绍
Ubuntu下安装并配置FastDFS

R脚本在Windows和Mac上的一点不兼容

Overview

今天在Mac上运行一个在Windows中处理过的R脚本,不能正常运行,错误信息指向明确。

问题描述

使用./H_gr500_tune.R运行,报下面的错误:

env: Rscript\r: No such file or directory

打开这个脚本,头几行内容如下:

#!/usr/bin/env Rscript
library(e1071)
library(plyr)
library(ROCR)

并没有发现异常,很明显是因为Windows下换行符\r\n引起的,但是使用sublime查找这个脚本,能查找\n,却查不到有\r,所以没办法在sublime中手动去掉\r
可能sublime在显示的时候自动把\r\n处理成了\n,但脚本中\r确实是存在的,所以尽管看不到,但运行脚本时会报错。

解决方案

使用下面的命令行,去掉\r,并生成一个新文件H_gr500_tune1.R

cat H_gr500_tune.R | tr -d "\r" > H_gr500_tune1.R

重新运行H_gr500_tune1.R,一切正常。

小结

Windows下拿过来的脚本,最好都check一下换行符的问题,不管会不会有问题,看一下总没错...