首页 > 代码库 > leofs 对象存储中一匹黑马
leofs 对象存储中一匹黑马
leofs centos 集群搭建和测试(二)
centos leofs 配置
leofs cluster 基本规划
Manager
IP: 10.39.1.23, 10.39.1.24
Name: manager_0@10.39.1.23, manager_1@10.39.1.24
Gateway
IP: 10.39.1.28
Name: gateway_0@10.39.1.28
Storage
IP: 10.39.1.25 .. 10.39.1.26 ... 10.39.1.27
Name: storage_01@10.39.1.25 .. storage_02@10.39.1.26...storage_02@10.39.1.27
LeoFS是一个非结构化的Web对象存储和高可用的，分布式的，最终一致的存储系统
10.39.1.23 上配置
yum install epel-release -y
yum install ansible -y
配置ansible
执行ansible
ansible leofs -m shell -a "yum install gcc gcc-c++ glibc-devel make ncurses-devel openssl-devel autoconf libuuid-devel cmake check check-devel wget curl git gcc* vim nc -y "
修改主机名以及hosts
10.39.1.23 leofs_01
10.39.1.24 leofs_02
10.39.1.25 storage_01
10.39.1.26 storage_02
10.39.1.27 storage_03
10.39.1.28 gw_01
生成key
ssh-keygen
ansible 推送key 到其他服务器
ansible all -m authorized_key -a "user=root key=‘{{ lookup(‘file‘,‘~/.ssh/id_rsa.pub‘)}}‘"
编辑所有服务器配置主机名
vim /etc/hostname
vim /etc/hosts
验证
ansible leofs -m shell -a "hostname"
ansible leofs -m shell -a "cat /etc/hosts"
下载leofs 包
http://leo-project.net/leofs/download.html
下载centos 7 1.3.2.1 最新rpm包
将rpm copy 到其他服务器
ansible all -m copy -a ‘src=http://www.mamicode.com/~/leofs-1.3.2.1-1.erl-18.3.el7.x86_64.rpm dest=/opt‘
ansible all -m shell -a ‘ls -l /opt‘
ansible all -m shell -a ‘cd /opt/; yum install leofs-1.3.2.1-1.erl-18.3.el7.x86_64.rpm -y‘
# 存储节点安装
ansible leofs_storage -m shell -a "yum --enablerepo=centosplus install kmod-xfs xfsprogs xfsprogs-devel -y"
查看安装目录是否全部安装
ansible all -m shell -a ‘ls -l /usr/local/leofs‘
配置leofs
10.39.1.23 配置manger0
manager0 中会配置一致性的级别和集群的id 以及名称
leofs 一致性级别配置参考
#leofs 集群一致性级别配置
相关资料参考 http://leo-project.net/leofs/docs/configuration/configuration_1.html#system-configuration-label
A reference consistency level
LevelConfiguration
Lown = 3, r = 1, w = 1, d = 1
Middlen = 3, [r = 1 | r = 2], w = 2, d = 2
Highn = 3, [r = 2 | r = 3], w = 3, d = 3
cp /usr/local/leofs/1.3.2.1/leo_manager_0/etc/leo_manager.conf /usr/local/leofs/1.3.2.1/leo_manager_0/etc/leo_manager.conf.bak
vim /usr/local/leofs/1.3.2.1/leo_manager_0/etc/leo_manager.conf
manager.partner = manager_1@10.39.1.24
system.dc_id = dc_1
system.cluster_id = leofs_cluster
consistency.num_of_replicas = 3
consistency.write = 1
consistency.read = 1
consistency.delete = 1
consistency.rack_aware_replicas = 0
nodename = manager_0@10.39.1.23
10.39.1.24 manger1 配置
cp /usr/local/leofs/1.3.2.1/leo_manager_1/etc/leo_manager.conf /usr/local/leofs/1.3.2.1/leo_manager_1/etc/leo_manager.conf.bak
vim /usr/local/leofs/1.3.2.1/leo_manager_1/etc/leo_manager.conf
manager.partner = manager_0@10.39.1.23
nodename = manager_1@10.39.1.24
10.39.1.25 storage_01 配置
返回到10.39.1.23 上使用ansible 对配置文件进行备份
ansible leofs_storage -m shell -a "cp /usr/local/leofs/1.3.2.1/leo_storage/etc/leo_storage.conf /usr/local/leofs/1.3.2.1/leo_storage/etc/leo_storage.conf.bak "
leofs 建议使用xfs，因为xfs I/O对大文件支持比较好
添加一块硬盘并且格式化为xfs 文件系统
fdisl /dev/vdc
n
p
w
mkfs.xfs /dev/vdc1
vim /usr/local/leofs/1.3.2.1/leo_storage/etc/leo_storage.conf
managers = [manager_0@10.39.1.23, manager_1@10.39.1.24]
obj_containers.path = [/data/leofs]
obj_containers.num_of_containers = [8]
#读写参数的一些设置
磁盘的一些设置
## Watchdog.DISK
##
## Is disk-watchdog enabled - default:false
watchdog.disk.is_enabled = false
## disk - raised error times
watchdog.disk.raised_error_times = 5
## disk - watch interval - default:1sec
watchdog.disk.interval = 10
## Threshold use(%) of a target disk‘s capacity - defalut:85%
watchdog.disk.threshold_disk_use = 85
## Threshold disk utilization(%) - defalut:90%
watchdog.disk.threshold_disk_util = 90
## Threshold disk read kb/sec - defalut:98304(KB) = 96MB
#watchdog.disk.threshold_disk_rkb = 98304
#131072kb=128MB
watchdog.disk.threshold_disk_rkb = 131072
## Threshold disk write kb/sec - defalut:98304(KB) = 96MB
#watchdog.disk.threshold_disk_wkb = 98304
#131072(kb)=128MB
watchdog.disk.threshold_disk_wkb = 131072
nodename = storage_01@10.39.1.25
配置也可以参考
leofs 存储默认可以设置多个
## e.g. Case of plural pathes
## obj_containers.path = [/var/leofs/avs/1, /var/leofs/avs/2]
## obj_containers.num_of_containers = [32, 64]
10.39.1.26 storage02 配置
mkfs.xfs /dev/vdc1
vim /usr/local/leofs/1.3.2.1/leo_storage/etc/leo_storage.conf
managers = [manager_0@10.39.1.23, manager_1@10.39.1.24]
obj_containers.path = [/data/leofs]
obj_containers.num_of_containers = [8]
nodename = storage_02@10.39.1.26
10.39.1.27 storage03 配置
mkfs.xfs /dev/vdc1
vim /usr/local/leofs/1.3.2.1/leo_storage/etc/leo_storage.conf
managers = [manager_0@10.39.1.23, manager_1@10.39.1.24]
obj_containers.path = [/data/leofs]
obj_containers.num_of_containers = [8]
nodename = storage_03@10.39.1.27
返回到10.39.1.23 上使用ansible 批量执行
ansible leofs_storage -m shell -a "mkdir /data/leofs -p "
ansible leofs_storage -m shell -a "ls /data/"
ansible leofs_storage -m shell -a "mount /dev/vdc1 /data/leofs"
ansible leofs_storage -m shell -a "df -h"
echo "/dev/vdc1 /data/leofs xfs noatime,nodiratime,osyncisdsync 0 0" >> /etc/fstab
ansible leofs_storage -m shell -a ‘echo "/dev/vdc1 /data/leofs xfs noatime,nodiratime,osyncisdsync 0 0" >> /etc/fstab‘
将storage01 02 03 重启
查看存储是否挂载成功
ansible leofs_storage -m shell -a "df -h"
gw_01 10.39.1.28 配置
网关配置协议 端口 网关的缓存大小
vim /usr/local/leofs/1.3.2.1/leo_gateway/etc/leo_gateway.conf
managers = [manager_0@10.39.1.23, manager_1@10.39.1.24]
protocol = s3
http.port = 8080
cache.cache_ram_capacity = 268435456
cache.cache_disc_capacity = 0
cache.cache_expire = 300
cache.cache_max_content_len = 1048576
nodename = gateway_01@10.39.1.28
# 网关配置线程池
## Large Object Handler - put worker pool size
large_object.put_worker_pool_size = 64
## Large Object Handler - put worker buffer size
large_object.put_worker_buffer_size = 32
## Memory cache capacity in bytes
cache.cache_ram_capacity = 0
## Disk cache capacity in bytes
cache.cache_disc_capacity = 0
## When the length of the object exceeds this value, store the object on disk
cache.cache_disc_threshold_len = 1048576
## Directory for the disk cache data
cache.cache_disc_dir_data = http://www.mamicode.com/cache/data
## Directory for the disk cache journal
cache.cache_disc_dir_journal = ./cache/journal
启动顺序
Order of server launch
Manager-master
Manager-slave
Storage nodes
Gateway(s)
10.39.1.23 manager0
/usr/local/leofs/1.3.2.1/leo_manager_0/bin/leo_manager start
/usr/local/leofs/1.3.2.1/leo_manager_0/bin/leo_manager ping
pong
10.39.1.24 manager1
/usr/local/leofs/1.3.2.1/leo_manager_1/bin/leo_manager start
/usr/local/leofs/1.3.2.1/leo_manager_1/bin/leo_manager ping
pong
10.39.1.25 storage01
/usr/local/leofs/1.3.2.1/leo_storage/bin/leo_storage start
/usr/local/leofs/1.3.2.1/leo_storage/bin/leo_storage ping
pong
10.39.1.26 storage02
/usr/local/leofs/1.3.2.1/leo_storage/bin/leo_storage start
/usr/local/leofs/1.3.2.1/leo_storage/bin/leo_storage ping
pong
10.39.1.27 storage03
/usr/local/leofs/1.3.2.1/leo_storage/bin/leo_storage start
/usr/local/leofs/1.3.2.1/leo_storage/bin/leo_storage ping
pong
查看存储的启动信息
ansible leofs_storage -m shell -a "ls -l /data/leofs"
10.39.1.26 | SUCCESS | rc=0 >>
总用量 12
drwxr-xr-x 2 root root 4096 3月 30 16:00 log
drwxr-xr-x 10 root root 4096 3月 30 15:24 metadata
drwxr-xr-x 2 root root 310 3月 30 15:18 object
drwxr-xr-x 2 root root 4096 3月 30 15:24 state
10.39.1.25 | SUCCESS | rc=0 >>
总用量 12
drwxr-xr-x 2 root root 4096 3月 30 16:00 log
drwxr-xr-x 10 root root 4096 3月 30 15:23 metadata
drwxr-xr-x 2 root root 310 3月 30 15:04 object
drwxr-xr-x 2 root root 4096 3月 30 15:23 state
10.39.1.27 | SUCCESS | rc=0 >>
总用量 4
drwxr-xr-x 2 root root 4096 3月 30 16:00 log
drwxr-xr-x 10 root root 246 3月 30 15:19 metadata
drwxr-xr-x 2 root root 310 3月 30 15:19 object
在10.39.1.23 上执行查看集群的状态
集群的状态查询是经过nmap 通信查询的， 需要在centos 下安装nc
ansible leofs -m shell -a "yum install nc -y "
/usr/local/leofs/1.3.2.1/leofs-adm status
/usr/local/leofs/1.3.2.1/leofs-adm status
[System Confiuration]
-----------------------------------+----------
Item | Value
-----------------------------------+----------
Basic/Consistency level
-----------------------------------+----------
system version | 1.3.2
cluster Id | leofs_cluster
DC Id | dc_1
Total replicas | 3
number of successes of R | 1
number of successes of W | 1
number of successes of D | 1
number of rack-awareness replicas | 0
ring size | 2^128
-----------------------------------+----------
Multi DC replication settings
-----------------------------------+----------
max number of joinable DCs | 2
number of replicas a DC | 1
-----------------------------------+----------
Manager RING hash
-----------------------------------+----------
current ring-hash |
previous ring-hash |
-----------------------------------+----------
[State of Node(s)]
-------+----------------------------+--------------+----------------+----------------+----------------------------
type | node | state | current ring | prev ring | updated at
-------+----------------------------+--------------+----------------+----------------+----------------------------
S | storage_01@10.39.1.25 | attached | | | 2017-03-30 15:24:56 +0800
S | storage_02@10.39.1.26 | attached | | | 2017-03-30 15:24:43 +0800
S | storage_03@10.39.1.27 | attached | | | 2017-03-30 15:19:22 +0800
-------+----------------------------+--------------+----------------+----------------+----------------------------
10.39.1.28
启动网关
/usr/local/leofs/1.3.2.1/leo_gateway/bin/leo_gateway start
/usr/local/leofs/1.3.2.1/leo_gateway/bin/leo_gateway ping
pong
10.39.1.23 leofs 存储系统启动
/usr/local/leofs/1.3.2.1/leofs-adm start
Generating RING...
Generated RING
OK 33% - storage_01@10.39.1.25
OK 67% - storage_03@10.39.1.27
OK 100% - storage_02@10.39.1.26
OK
/usr/local/leofs/1.3.2.1/leofs-adm status
[System Confiuration]
-----------------------------------+----------
Item | Value
-----------------------------------+----------
Basic/Consistency level
-----------------------------------+----------
system version | 1.3.2
cluster Id | leofs_cluster
DC Id | dc_1
Total replicas | 3
number of successes of R | 1
number of successes of W | 1
number of successes of D | 1
number of rack-awareness replicas | 0
ring size | 2^128
-----------------------------------+----------
Multi DC replication settings
-----------------------------------+----------
max number of joinable DCs | 2
number of replicas a DC | 1
-----------------------------------+----------
Manager RING hash
-----------------------------------+----------
current ring-hash |
previous ring-hash |
-----------------------------------+----------
[State of Node(s)]
-------+----------------------------+--------------+----------------+----------------+----------------------------
type | node | state | current ring | prev ring | updated at
-------+----------------------------+--------------+----------------+----------------+----------------------------
S | storage_01@10.39.1.25 | running | 79e0dbc4 | 79e0dbc4 | 2017-03-30 15:29:17 +0800
S | storage_02@10.39.1.26 | running | 79e0dbc4 | 79e0dbc4 | 2017-03-30 15:29:17 +0800
S | storage_03@10.39.1.27 | running | 79e0dbc4 | 79e0dbc4 | 2017-03-30 15:29:17 +0800
G | gateway_01@10.39.1.28 | running | 79e0dbc4 | 79e0dbc4 | 2017-03-30 15:29:18 +0800
-------+----------------------------+--------------+----------------+----------------+----------------------------
leofs 对象存储的配置和查询
s3-api 命令使用
http://leo-project.net/leofs/docs/admin_guide/admin_guide_8.html
用户查询
get-users
/usr/local/leofs/1.3.2.1/leofs-adm get-users
user_id | role_id | access_key_id | created_at
------------+---------+------------------------+---------------------------
_test_leofs | 9 | 05236 | 2017-03-30 15:03:54 +0800
删除默认的用户
delete-user <user-id>
/usr/local/leofs/1.3.2.1/leofs-adm delete-user _test_leofs
OK
创建用户
create-user <user-id> <password>
/usr/local/leofs/1.3.2.1/leofs-adm create-user test test
access-key-id: 919ca38e3fb34085b94a
secret-access-key: 387d7f32546982131e41355e1adbcae1a9b08bec
/usr/local/leofs/1.3.2.1/leofs-adm get-users
user_id | role_id | access_key_id | created_at
--------+---------+------------------------+---------------------------
test | 1 | 919ca38e3fb34085b94a | 2017-03-30 15:45:41 +0800
endpoint
/usr/local/leofs/1.3.2.1/leofs-adm add-endpoint 10.39.1.28
OK
/usr/local/leofs/1.3.2.1/leofs-adm get-endpoints
endpoint | created at
-----------------+---------------------------
10.39.1.28 | 2017-03-30 15:49:14 +0800
localhost | 2017-03-30 15:03:54 +0800
s3.amazonaws.com | 2017-03-30 15:03:54 +0800
add-bucket <bcuket> <access-key-id>
/usr/local/leofs/1.3.2.1/leofs-adm add-bucket abc 919ca38e3fb34085b94a
OK
/usr/local/leofs/1.3.2.1/leofs-adm get-buckets
cluster id | bucket | owner | permissions | redundancy method | created at
--------------+----------+--------+------------------+------------------------------+---------------------------
leofs_cluster | abc | test | Me(full_control) | copy, {n:3, w:1, r:1, d:1} | 2017-03-30 15:51:43 +0800
get-bucket <access-key-id>
权限
/usr/local/leofs/1.3.2.1/leofs-adm update-acl abc 919ca38e3fb34085b94a public-read-write
OK
查看存储节点的磁盘信息
/usr/local/leofs/1.3.2.1/leofs-adm du detail storage_01@10.39.1.25
[du(storage stats)]
file path: /data/leofs/object/0.avs
active number of objects: 0
total number of objects: 0
active size of objects: 0
total size of objects: 0
ratio of active size: 0%
last compaction start: ____-__-__ __:__:__
last compaction end: ____-__-__ __:__:__
duration: 0s
result: ‘‘
file path: /data/leofs/object/1.avs
active number of objects: 0
total number of objects: 0
active size of objects: 0
total size of objects: 0
ratio of active size: 0%
last compaction start: ____-__-__ __:__:__
last compaction end: ____-__-__ __:__:__
duration: 0s
result: ‘‘
file path: /data/leofs/object/2.avs
active number of objects: 0
total number of objects: 0
active size of objects: 0
total size of objects: 0
ratio of active size: 0%
last compaction start: ____-__-__ __:__:__
last compaction end: ____-__-__ __:__:__
duration: 0s
result: ‘‘
file path: /data/leofs/object/3.avs
active number of objects: 0
total number of objects: 0
active size of objects: 0
total size of objects: 0
ratio of active size: 0%
last compaction start: ____-__-__ __:__:__
last compaction end: ____-__-__ __:__:__
duration: 0s
result: ‘‘
file path: /data/leofs/object/4.avs
active number of objects: 0
total number of objects: 0
active size of objects: 0
total size of objects: 0
ratio of active size: 0%
last compaction start: ____-__-__ __:__:__
last compaction end: ____-__-__ __:__:__
duration: 0s
result: ‘‘
file path: /data/leofs/object/5.avs
active number of objects: 0
total number of objects: 0
active size of objects: 0
total size of objects: 0
ratio of active size: 0%
last compaction start: ____-__-__ __:__:__
last compaction end: ____-__-__ __:__:__
duration: 0s
result: ‘‘
file path: /data/leofs/object/6.avs
active number of objects: 0
total number of objects: 0
active size of objects: 0
total size of objects: 0
ratio of active size: 0%
last compaction start: ____-__-__ __:__:__
last compaction end: ____-__-__ __:__:__
duration: 0s
result: ‘‘
file path: /data/leofs/object/7.avs
active number of objects: 0
total number of objects: 0
active size of objects: 0
total size of objects: 0
ratio of active size: 0%
last compaction start: ____-__-__ __:__:__
last compaction end: ____-__-__ __:__:__
duration: 0s
result: ‘‘
/usr/local/leofs/1.3.2.1/leofs-adm status storage_01@10.39.1.25
--------------------------------------+--------------------------------------
Item | Value
--------------------------------------+--------------------------------------
Config-1: basic
--------------------------------------+--------------------------------------
version | 1.3.1
number of vnodes | 168
object containers | - path:[/data/leofs], # of containers:8
log directory | ./log/erlang
log level | info
--------------------------------------+--------------------------------------
Config-2: watchdog
--------------------------------------+--------------------------------------
[rex(rpc-proc)] |
check interval(s) | 10
threshold mem capacity | 33554432
--------------------------------------+--------------------------------------
[cpu] |
enabled/disabled | disabled
check interval(s) | 10
threshold cpu load avg | 5.0
threshold cpu util(%) | 100
--------------------------------------+--------------------------------------
[disk] |
enabled/disalbed | disabled
check interval(s) | 10
threshold disk use(%) | 85
threshold disk util(%) | 90
threshold rkb(kb) | 131072
threshold wkb(kb) | 131072
--------------------------------------+--------------------------------------
Config-3: message-queue
--------------------------------------+--------------------------------------
number of procs/mq | 8
number of batch-procs of msgs | max:3000, regular:1600
interval between batch-procs (ms) | max:3000, regular:500
--------------------------------------+--------------------------------------
Config-4: autonomic operation
--------------------------------------+--------------------------------------
[auto-compaction] |
enabled/disabled | disabled
warning active size ratio (%) | 70
threshold active size ratio (%) | 60
number of parallel procs | 1
exec interval | 3600
--------------------------------------+--------------------------------------
Config-5: data-compaction
--------------------------------------+--------------------------------------
limit of number of compaction procs | 4
number of batch-procs of objs | max:1500, regular:1000
interval between batch-procs (ms) | max:3000, regular:500
--------------------------------------+--------------------------------------
Status-1: RING hash
--------------------------------------+--------------------------------------
current ring hash | 79e0dbc4
previous ring hash | 79e0dbc4
--------------------------------------+--------------------------------------
Status-2: Erlang VM
--------------------------------------+--------------------------------------
vm version | 7.3.1.2
total mem usage | 37035408
system mem usage | 20786136
procs mem usage | 16258552
ets mem usage | 5229136
procs | 560/1048576
kernel_poll | true
thread_pool_size | 32
--------------------------------------+--------------------------------------
Status-3: Number of messages in MQ
--------------------------------------+--------------------------------------
replication messages | 0
vnode-sync messages | 0
rebalance messages | 0
--------------------------------------+--------------------------------------
性能压测
leofs 性能测试工具 basho_bench
https://github.com/basho/basho_bench
10.39.1.23 安装
安装依赖
1. Install libatomic
##
$ wget http://www.ivmaisoft.com/_bin/atomic_ops/libatomic_ops-7.4.4.tar.gz
$ tar xzvf libatomic_ops-7.4.4.tar.gz
$ cd libatomic_ops-7.4.4
$ ./configure --prefix=/usr/local
$ make
$ sudo make install
##
## 2. Install Erlang (18.3)
##
$ wget http://erlang.org/download/otp_src_18.3.tar.gz
$ tar xzf otp_src_18.3.tar.gz
$ cd otp_src_18.3
$ ./configure --prefix=/usr/local/erlang/18.3 \
--enable-smp-support \
--enable-m64-build \
--enable-halfword-emulator \
--enable-kernel-poll \
--without-javac \
--disable-native-libs \
--disable-hipe \
--disable-sctp \
--enable-threads \
--with-libatomic_ops=/usr/local
$ make
$ sudo make install
##
## 3. Set PATH
##
$ vi ~/.profile
## append the follows:
export ERL_HOME=/usr/local/erlang/18.3
export PATH=$PATH:$ERL_HOME/bin
$ source ~/.profile
安装性能测试工具
git clone https://github.com/leo-project/basho_bench.git
cd basho_bench/
make all
# make all 的时候使用的git 有时候拉取依赖会报错，使用https 就好
批量替换git deps 目录下的所有的rebar.confg 配置文件git 修改为https
find deps -name rebar.config -type f -exec sed -i ‘s/git:\/\//https:\/\//g‘ {} +
make all
性能压测测试文件
16kb 文件压测
vim 16file.conf
{mode, max}.
{duration, 10}.
{concurrent, 50}.
{driver, basho_bench_driver_leofs}.
{code_paths, ["deps/ibrowse"]}.
{http_raw_ips, ["10.39.1.28"]}. %% able to set plural nodes
{http_raw_port, 8080}. %% default: 8080
{http_raw_path, "/abc"}.
%% {http_raw_path, "/${BUCKET}"}.
{key_generator, {partitioned_sequential_int, 660000}}. %% 请求的次数
{value_generator, {fixed_bin, 16384}}. %% 16KB
{operations, [{put,1}]}. %% PUT:100%
%%{operations, [{put,1}, {get, 4}]}. %% PUT:20%, GET:80%
{check_integrity, false}.
660000x16kb=10GB
安装nmon 收集信息
ansible leofs -m shell -a "yum install wget https://raw.githubusercontent.com/hambuergaer/nmon-packages/master/nmon-rhel7-1.0-1.x86_64.rpm -y "
ansible leofs -m shell -a "mkdir /nmon "
10 分钟
nmon 使用参数
-s 表示秒级采集一次数据
-c 表示采集收的次数
-m 表示生成数据文件的路径
-f 表示生成数据文件名中有时间
nmon -f -s 1 -c 360 -m /nmon
一秒采集一次，总共采集6分钟，1x360/3600=6分钟
先检查一下磁盘的大小
[root@leofs_01 basho_bench]# ansible leofs_storage -m shell -a "du -sh /data/leofs "
10.39.1.26 | SUCCESS | rc=0 >>
332K/data/leofs
10.39.1.25 | SUCCESS | rc=0 >>
332K/data/leofs
10.39.1.27 | SUCCESS | rc=0 >>
132K/data/leofs
同时写入9.7GB 16KB大小的文件 ，用时5分钟
用时5分钟
ansible leofs_storage -m shell -a "du -sh /data/leofs "
10.39.1.26 | SUCCESS | rc=0 >>
9.7G/data/leofs
10.39.1.25 | SUCCESS | rc=0 >>
9.7G/data/leofs
10.39.1.27 | SUCCESS | rc=0 >>
9.7G/data/leofs
把网关的线程池打开设置为64
# 网关配置线程池
## Large Object Handler - put worker pool size
large_object.put_worker_pool_size = 16
## Large Object Handler - put worker buffer size
large_object.put_worker_buffer_size = 32
## Memory cache capacity in bytes
cache.cache_ram_capacity = 0
## Disk cache capacity in bytes
cache.cache_disc_capacity = 0
## When the length of the object exceeds this value, store the object on disk
cache.cache_disc_threshold_len = 1048576
## Directory for the disk cache data
cache.cache_disc_dir_data = http://www.mamicode.com/cache/data
## Directory for the disk cache journal
cache.cache_disc_dir_journal = ./cache/journal
10万 文件用时一分钟 使用比例如下
4096.. 8192: 15%
8192.. 16384: 25%
16384.. 32768: 23%
32768.. 65536: 22%
65536.. 131072: 15%
{mode, max}.
{duration, 1000}.
{concurrent, 64}.
{driver, basho_bench_driver_leofs}.
{code_paths, ["deps/ibrowse"]}.
{http_raw_ips, ["10.39.1.28"]}.
{http_raw_port, 8080}.
{http_raw_path, "/abc"}.
{retry_on_overload, true}.
{key_generator, {partitioned_sequential_int, 1000000}}.
{value_generator, {fixed_bin, 262144}}.
{operations, [{put,1}]}.
{value_generator_source_size, 1048576}.
{http_raw_request_timeout, 30000}. % 30seconds
{value_size_groups, [{15, 4096, 8192},{25, 8192, 16384}, {23, 16384, 32768}, {22, 32768, 65536}, {15, 65536, 131072}]}.
查看存储空间大小
ansible leofs_storage -m shell -a "du -sh /data/leofs "
10.39.1.25 | SUCCESS | rc=0 >>
132K/data/leofs
10.39.1.26 | SUCCESS | rc=0 >>
132K/data/leofs
10.39.1.27 | SUCCESS | rc=0 >>
132K/data/leofs
[root@leofs_01 basho_bench]# ansible leofs_storage -m shell -a "ls /data/leofs "
10.39.1.26 | SUCCESS | rc=0 >>
log
metadata
object
10.39.1.27 | SUCCESS | rc=0 >>
log
metadata
object
10.39.1.25 | SUCCESS | rc=0 >>
log
metadata
object
ansible leofs_storage -m shell -a "du -sh /data/leofs "
10.39.1.25 | SUCCESS | rc=0 >>
4.1G/data/leofs
10.39.1.26 | SUCCESS | rc=0 >>
4.1G/data/leofs
10.39.1.27 | SUCCESS | rc=0 >>
4.1G/data/leofs
ansible leofs_storage -m shell -a "ls /data/leofs "
10.39.1.26 | SUCCESS | rc=0 >>
log
metadata
object
10.39.1.27 | SUCCESS | rc=0 >>
log
metadata
object
10.39.1.25 | SUCCESS | rc=0 >>
log
metadata
object
state
ansible leofs_storage -m shell -a "nmon -f -s 1 -c 1440 -m /nmon "
采集24分钟内的数据
100万 5%写 95% 读
{mode, max}.
{duration, 30}.
{concurrent, 64}.
{driver, basho_bench_driver_leofs}.
{code_paths, ["deps/ibrowse"]}.
{http_raw_ips, ["192.168.100.35"]}.
{http_raw_port, 8080}.
{http_raw_path, "/test"}.
{retry_on_overload, true}.
{key_generator, {uniform_int,1000000}}.
{value_generator, {fixed_bin, 262144}}.
{operations, [{put,5}, {get, 95}]}.
{value_generator_source_size, 1048576}.
{http_raw_request_timeout, 30000}. % 30seconds
{value_size_groups, [{15, 4096, 8192},{25, 8192, 16384}, {23, 16384, 32768}, {22, 32768, 65536}, {15, 65536, 131072}]}.
10.39.1.23 安装LeoCenter
git clone https://github.com/leo-project/leo_center.git
yum install ruby-devel -y
cd leo_center/
gem install bundler
bundle install
修改配置
config.yml
:managers:
- "localhost:10020" # master
- "localhost:10021" # slave
:credential:
:access_key_id: "YOUR_ACCESS_KEY_ID"
:secret_access_key: "YOUR_SECRET_ACCESS_KEY"
启动服务
thin start -a ${HOST} -p ${PORT} > /dev/null 2>&1 &
创建管理员
You need to create an administrator user from LeoFS-Manager’s console.
$ leofs-adm create-user leo_admin password
access-key-id: ab96d56258e0e9d3621a
secret-access-key: 5c3d9c188d3e4c4372e414dbd325da86ecaa8068
$ leofs-adm update-user-role leo_admin 9
OK
leofs-adm create-user leo_admin password
access-key-id: 8d9de6fdfb35f837e9ed
secret-access-key: 7dcc2631493865c7fb7ec7f96dda627f1cbb21eb
leofs-adm update-user-role leo_admin 9
OK
[root@leofs_01 leo_center]# leofs-adm get-users
user_id | role_id | access_key_id | created_at
----------+---------+------------------------+---------------------------
leo_admin | 9 | 8d9de6fdfb35f837e9ed | 2017-03-31 11:40:14 +0800
test | 1 | 919ca38e3fb34085b94a | 2017-03-30 15:45:41 +0800
经过测试。
删除用户buckets 是存在的，只有当buckets 删除之后数据才能真正删除
本文出自 “小菜鸟” 博客，请务必保留此出处http://xiaocainiaox.blog.51cto.com/4484443/1913296
leofs 对象存储中一匹黑马