【FastDFS分布式文件系统之一】：搭建、部署、配置

        最近要提高小文件（100KB左右）的存储性能（主要是写入性能），想到了使用FastDFS因此需要对其性能做一个简单的测试。

1.安装FastDFS

cd /opt && mkdir fastdfs && cd fastdfsgit clone https://github.com/happyfish100/fastdfs.gitgit clone https://github.com/happyfish100/libfastcommon.gitcd /opt/fastdfs/libfastcommon./make.sh && ./make.sh installcd /opt/fastdfs/fastdfs./make.sh && ./make.sh install

安装成功后，FastDFS安装在/usr/bin中。配置文件在/etc/fdfs中：

2.修改配置文件

配置文件放在/etc/fdfs下面：

client配置文件：/etc/fdfs/client.conf.sample

storage配置文件：/etc/fdfs/storage.com.sample

tracker配置文件：/etc/fdfs/tracker.com.sample

cd /etc/fdfsmv client.conf.sample client.confmv storage.conf.sample storage.confmv tracker.conf.sample tracker.conf

client：node1

tracker server：node2 / node3

group1：storage server1(node4) /storage server2(node5) /storage server3(node6)

group2：storage server4(node7) /storage server5(node6) /storage server6(node8)

修改client.conf：

base_path=/mnt/xfsd/fastdfs/clienttracker_server=node2:22122tracker_server=node3:22122#其他都采用默认

其中base_path是客户端的数据和日志写入的目录，这个目录必须指定且提前创建好。

修改storage.conf：

base_path=/mnt/xfsd/fastdfs/storagemax_connections=2048accept_threads=4work_threads=8tracker_server=node2:22122tracker_server=node3:22122log_level=info#其他都采用默认

max_connections：并发最大请求数，默认256，如果设置的比较小，那么如果并发请求达到300以上，就会造成失败，如果设置的太大（20480）又会造成storage server启动失败，不知道为什么。具体的参数配置可以见配置文件的注释。

修改tracker.conf：

base_path=/mnt/xfsd/fastdfs/trackermax_connections=2048accept_threads=4work_threads=8log_level=infouse_trunk_file = true

值得一提的是use_trunk_file来进行小文件的合并，默认合并成为64MB的块。

3.启动服务

服务的启动，重启，状态查询等。

/usr/bin/fdfs_trackerd /etc/fdfs/tracker.conf/usr/bin/fdfs_storaged /etc/fdfs/storage.confservice fdfs_storaged statusservice fdfs_storaged startservice fdfs_storaged stopservice fdfs_storaged restartservice fdfs_trackerd statusservice fdfs_trackerd startservice fdfs_trackerd restartservice fdfs_trackerd stop

4.检查服务

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

输出如下信息：

[2016-03-08 17:40:36] DEBUG - base_path=/mnt/xfsd/fastdfs/storage, connect_timeout=30, network_timeout=60, tracker_server_count=2, 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: 0server_count=2, server_index=1tracker server is XXXXXX:22122group count: 2Group 1:group name = group1disk total space = 6654860 MBdisk free space = 6430591 MBtrunk free space = 127 MBstorage server count = 3active server count = 3storage server port = 23000storage HTTP port = 8888store path count = 1subdir count per path = 256current write server index = 0current trunk file id = 3481Storage 1:id = XXX.119ip_addr = XXX.119  ACTIVEhttp domain = version = 5.08join time = 2016-03-07 21:46:38up time = 2016-03-07 21:46:38total storage = 6654860 MBfree storage = 6431077 MBupload priority = 10store_path_count = 1subdir_count_per_path = 256storage_port = 23000storage_http_port = 8888\Storage 2:    ……    Storage 3:……Group 2:group name = group2disk total space = 6654860 MBdisk free space = 6429913 MBtrunk free space = 152 MBstorage server count = 3active server count = 3storage server port = 23000storage HTTP port = 8888store path count = 1subdir count per path = 256current write server index = 2current trunk file id = 3501Storage 1:id = XXX.122ip_addr = XXX.122  ACTIVEhttp domain = version = 5.08join time = 2016-03-07 21:50:57up time = 2016-03-07 21:50:57total storage = 6654860 MBfree storage = 6429917 MBupload priority = 10store_path_count = 1subdir_count_per_path = 256storage_port = 23000storage_http_port = 8888current_write_path = 0source storage id = if_trunk_server = 1Storage 2:    ……    Storage 3:……

5.数据操作

在client端执行测试：

上传upload操作：

/usr/bin/fdfs_test /etc/fdfs/client.conf upload /data/files/bak/smallfile0

下载download操作：

6.重新部署方法

      由于测试需要，将replica 2变成replica 3，那么只需要将tracer server以及storage server端的base_path下面的client,storage,tracker文件夹都删除掉，然后重新创建文件夹，重启各个server服务就好了，无需卸载，重新编译安装等。

Author：忆之独秀

Email：[e-mail]

注明出处：http://blog.csdn.net/lavorange/article/details/50814086