部署Ceph存储集群及块设备测试

509次阅读

共计 2908 个字符，预计需要花费 8 分钟才能阅读完成。

集群环境

部署 Ceph 存储集群及块设备测试

添加 ceph.repo

wget -O /etc/yum.repos.d/ceph.repo https://raw.githubusercontent.com/aishangwei/ceph-demo/master/ceph-deploy/ceph.repo
yum makecache

配置 NTP

yum -y install ntpdate ntp
ntpdate cn.ntp.org.cn
systemctl restart ntpd ntpdate;systemctl enable ntpd ntpdate

创建用户和 ssh 免密登录

useradd ceph-admin
echo "ceph-admin"|passwd --stdin ceph-admin
echo "ceph-admin ALL = (root) NOPASSWD:ALL" | sudo tee /etc/sudoers.d/ceph-admin
sudo chmod 0440 /etc/sudoers.d/ceph-admin

 配置 host 解析
cat >>/etc/hosts<<EOF
10.1.10.201 ceph01
10.1.10.202 ceph02
10.1.10.203 ceph03
EOF

配置 sudo 不需要 tty

sed -i 's/Default requiretty/#Default requiretty/' /etc/sudoers

配置免密登录

su - ceph-admin
ssh-keygen
ssh-copy-id ceph-admin@ceph01
ssh-copy-id ceph-admin@ceph02
ssh-copy-id ceph-admin@ceph03

安装 ceph-deploy

sudo yum install -y ceph-deploy Python-pip

部署节点

mkdir my-cluster;cd my-cluster
ceph-deploy new ceph01 ceph02 ceph03

编辑 ceph.conf 配置文件

echo >>/home/ceph-admin/my-cluster/ceph.conf<<EOF
public network = 10.1.10.0/16
cluster network = 10.1.10.0/16
EOF

安装 ceph 包（代替 ceph-deploy install node1 node2, 下面命令需要在每台 node 上安装）

sudo wget -O /etc/yum.repos.d/epel.repo http://mirrors.aliyun.com/repo/epel-7.repo
rpm --import /etc/pki/rpm-gpg/RPM-GPG-KEY-EPEL-7
sudo yum install -y ceph ceph-radosgw

配置初始 monitor(s), 收集所有密钥

ceph-deploy mon create-initial
ls -l *.keyring

把配置信息拷贝到各节点

ceph-deploy admin ceph01 ceph02 ceph03

配置 osd

su - ceph-admin
cd /home/my-cluster

for dev in /dev/sdb /dev/sdc /dev/sdd
do
ceph-deploy disk zap ceph01 $dev
ceph-deploy osd create ceph01 --data $dev
ceph-deploy disk zap ceph02 $dev
ceph-deploy osd create ceph02 --data $dev
ceph-deploy disk zap ceph03 $dev
ceph-deploy osd create ceph03 --data $dev
done

部署 mgr,Luminous 版以后才需要部署

ceph-deploy mgr create ceph01 ceph02 ceph03

开启 dashboard 模块

sudo chown -R ceph-admin /etc/ceph/
ceph mgr module enable dashboard
netstat -lntup|grep 7000

http://10.1.10.201:7000

检查是否复合块设备环境要求

uname -r
modprobe rbd
echo $?

创建池和块设备

ceph osd lspools
ceph osd pool create rbd 128

确定 pg_num 取值是强制性的，因为不能自动计算，下面是几个常用的值

少于 5 个 OSD 时,pg_num 设置为 128
OSD 数量在 5 到 10 个时，pg_num 设置为 512
OSD 数量在 10 到 50 个时，pg_num 设置为 4096
OSD 数量大于 50 时，理解权衡方法、以及如何自己计算 pg_num 取值

客户端创建块设备

rbd create rbd1 --size 1G --image-feature layering --name client.admin

映射块设备

rbd map --image rbd1 --name client.admin

创建文件系统并挂载

fdisk -l /dev/rbd0
mkfs.xfs /dev/rbd0
mkdir /mnt/ceph-disk1
mount /dev/rbd0 /mnt/ceph-disk1
df -h /mnt/ceph-disk1

写入数据测试

dd if=/dev/zero of=/mnt/ceph-disk1/file1 count=100 bs=1M

安装 fio 压测软件

yum install zlib-devel -y
yum install ceph-devel -y
git clone git://git.kernel.dk/fio.git
cd fio/
./configure
make;make install

测试磁盘性能

fio -direct=1 -iodepth=1 -rw=read -ioengine=libaio -bs=2k -size=100G -numjobs=128 -runtime=30 -group_reporting - 
filename=/dev/rbd0 -name=readiops
fio -direct=1 -iodepth=1 -rw=write -ioengine=libaio -bs=2k -size=100G -numjobs=128 -runtime=30 -group_reporting - 
filename=/dev/rbd0 -name=writeiops
fio -direct=1 -iodepth=1 -rw=randread -ioengine=libaio -bs=2k -size=100G -numjobs=128 -runtime=30 -group_reporting - 
filename=/dev/rbd0 -name=randreadiops
fio -direct=1 -iodepth=1 -rw=randwrite -ioengine=libaio -bs=2k -size=100G -numjobs=128 -runtime=30 -group_reporting - 
filename=/dev/rbd0 -name=randwriteiops