Ceph安装部署文档
目录
一:简介... 1
二:部署环境介绍... 1
三:集群配置准备工作... 2
四:安装ceph软件包... 6
五:安装ceph对象网关... 9
六、搭建ceph集群... 10
6.1、配置mon节点... 10
6.2:添加osd节点... 13
6.2.1、添加第一块osd节点... 13
6.3:添加元数据服务器... 17
6.3.1、添加第一个元数据服务器... 17
七:安装client端RBD、cephFS挂载演示... 18
7.1:客户端内核要求、安装装准备... 18
7.2、RBD方式挂载... 19
7.3、cephFS挂载... 20
八:检查ceph集群状态常用命令... 23
8.1、磁盘压力测试软件推荐... 23
九:故障排查... 23
一:简介
Ceph 生态系统架构可以划分为四部分:
1. Clients:客户端(数据用户)
2. mds:Metadata servercluster,元数据服务器(缓存和同步分布式元数据)
3. osd:Object storagecluster,对象存储集群(将数据和元数据作为对象存储,执行其它关键职能)
4. mon:Cluster monitors,集群监视器(执行监视功能)
二:部署环境介绍
主机名 |
公网IP(eth0) |
运行服务 |
cpu |
内存 |
操作系统 |
磁盘 |
备注 |
client |
172.16.4.17 |
Client |
cpu: e5520*2 |
4g*4 |
CentOS6.6 |
||
node01 |
172.16.2.27 |
mon、mds、osd |
cpu: e5520*2 |
4g*4 |
CentOS6.6 |
1T sas 7500转*4 |
|
node02 |
172.16.2.28 |
mon、mds、osd |
cpu: e5520*2 |
4g*4 |
CentOS6.6 |
1T sas 7500转*4 |
|
node03 |
172.16.2.29 |
mon、mds、osd |
cpu: e5520*2 |
4g*4 |
CentOS6.6 |
1T sas 7500转*4 |
三:集群配置准备工作
node01节点执行
3.1:创建工作文件夹,建立ip地址list,为文件传输做准备
mkdir /workspace/
[root@node01 ~]#cd /workspace/
[root@node01workspace]# vi cephlist.txt 主机列表
client
node01
node02
node03
主机ssh登陆证书
ssh-keygen
cd .ssh/
ssh-copy-id 127.1
将配置完成的证书传输到其它服务器
sync –avp * 172.17.2.27:/root/.ssh/
sync –avp * 172.17.2.28:/root/.ssh/
sync –avp * 172.17.2.29:/root/.ssh/
3.2:系统基础设置,建议整理为脚本,批量传输到所有集群主机执行
3.2.1.修改SELINUX模式
sed -i's/SELINUX=enforcing/SELINUX=disabled/' /etc/sysconfig/selinux
sed -i's/SELINUX=enforcing/SELINUX=disabled/' /etc/selinux/config
3.2.2.set max userprocesses
sed -i's/1024/102400/' /etc/security/limits.d/90-nproc.conf
echo "* -sigpending 256612" >> /etc/security/limits.d/90-nproc.conf
3.2.3.set ulimit
echo "ulimit-SHn 102400" >> /etc/rc.local
cat >>/etc/security/limits.conf << EOF
* soft nofile 102400
* hard nofile 102400
* soft nproc 102400
* - sigpending256612
* hard nproc 102400
EOF
#执行profile
source/etc/profile
3.3: 修改host文件
[root@node01 ~]# vi /etc/hosts
追加内容如下:
172.16.4.17 client
172.16.2.27 node01
172.16.2.28 node02
172.16.2.29 node03
172.16.4.240 cephNaNmind.com
将host文件传输到其它主机
for ip in $(cat/workspace/cephlist.txt);do echo -----$ip-----;rsync -avp /etc/hosts $ip:/etc/;done
3.4:安装centos的yum源软件包 =>全部节点安装
3.4.1:升级基础包
rpm -Uvhhttp://mirrors.sohu.com/centos/6.6/os/x86_64/Packages/python-iniparse-0.3.1-2.1.el6.noarch.rpm
rpm -Uvh http://mirrors.sohu.com/centos/6.6/os/x86_64/Packages/yum-metadata-parser-1.1.2-16.el6.x86_64.rpm
rpm -Uvhhttp://mirrors.sohu.com/centos/6.6/os/x86_64/Packages/yum-3.2.29-60.el6.centos.noarch.rpm
rpm -Uvhhttp://mirrors.sohu.com/centos/6.6/os/x86_64/Packages/yum-plugin-fastestmirror-1.1.30-30.el6.noarch.rpm
3.4.2:配置一个国内yum源
vi/etc/yum.repos.d/my.repo[k1]
详见批注
3.4.3:添加时间同步定时任务
vi /etc/crontab
3.4.4:时间同步,此处是我们的内部时间源
20 */1 * * * root/usr/sbin/ntpdate 172.16.5.100 && /sbin/hwclock –w
for ip in $(cat /workspace/cephlist.txt);doecho -----$ip-----;rsync -avp --delete /etc/crontab $ip:/etc/;done
4:5:传输yum源文件到其它节点服务器
--delete 删除那些DST中SRC没有的文件
for ip in $(cat/workspace/cephlist.txt);do echo -----$ip-----;rsync -avp --delete/etc/yum.repos.d $ip:/etc/;done
3.5、更新yum源 (所有节点执行)
Screen
yum clean all&&yum update –y
3.6、添加ceph yum源或者自制yum源;
在所有节点安装ceph的所有的yum源 (所有节点执行)
3.6.1、安装软件包key
(1)、release.asckey
rpm --import'https://ceph.com/git/?p=ceph.git;a=blob_plain;f=keys/release.asc'
(2)、autobuild.asckey
rpm --import'https://ceph.com/git/?p=ceph.git;a=blob_plain;f=keys/autobuild.asc'
3.6.2、安装ceph附加包源ceph-extras.repo,设置priority=2,确保新的包(如qemu)优先级比标准包的高。
vi/etc/yum.repos.d/ceph-extras.repo[k2] [k3]
详见附件
#制作私有yum源下载软件包地址
wget -S -c -r -np-L http://ceph.com/packages/ceph-extras/rpm/centos6/
3.6.3、安装ceph源
vi/etc/yum.repos.d/ceph.repo[k4] [k5]
##制作私有yum源下载软件包地址
wget -S -c -r -np -L http://ceph.com/rpm/rhel6/
3.6.4、安装ceph的apache yum源
vi/etc/yum.repos.d/ceph-apache.repo[k6]
##制作私有yum源下载软件包地址
wget -S -c -r -np-L http://gitbuilder.ceph.com/apache2-rpm-centos6-x86_64-basic/ref/master/
3.6.5、添加ceph-fastcgi.repofile to /etc/yum.repos.d
vi/etc/yum.repos.d/mod_fastcgi.repo[k7]
详见批注
##制作私有yum源下载软件包地址
wget -S -c -r -np-L http://gitbuilder.ceph.com/mod_fastcgi-rpm-centos6-x86_64-basic/ref/master/
3.7:如果制作内网yum源,完成上面给出的下载链接后,进行如下配置:
3.7.1:删除无用文件
cd/usr/share/nginx/html
find ./ -nameindex.html* >1.txt
for i in $(cat1.txt);do rm -rf $i;done
find ./ -nameindex.html*
3.7.2:调整目录结构
下载至本地的软件包,目录结构根据实际情况作调整,
3.7.3:添加、修改nginx配置:
location / {
root /usr/share/nginx/html;
autoindex on;
3.7.4、修改yum源,repo文件,将yum请求的地址改为私有地址;
例如:
vi/etc/yum.repos.d/ceph-extras.repo[k8]
[ceph-extras-source]
name=Ceph ExtrasSources
baseurl=http://172.16.2.240/ceph-extras/packages/ceph-extras/rpm/centos6/SRPMS
enabled=1
priority=2
gpgcheck=0
type=rpm-md
gpgkey=https://ceph.com/git/?p=ceph.git;a=blob_plain;f=keys/release.asc
3.8、安装国内epel yum源
rpm -Uvhhttp://mirrors.sohu.com/fedora-epel/6/x86_64/epel-release-6-8.noarch.rpm
备:
rpm -Uvhhttp://dl.fedoraproject.org/pub/epel/6/x86_64/epel-release-6-8.noarch.rpm
官方推荐rpm安装ceph,安装下面插件:
3.9:安装yum-plugin-priorities.
yum install yum-plugin-priorities -y
yum update –y
3.9.1:确认配置文件内容
cat/etc/yum/pluginconf.d/priorities.conf
[main]
enabled = 1
[k1][base]
name=CentOS-6 -Base - 163.com
baseurl=http://mirrors.163.com/centos/6/os/$basearch/
#mirrorlist=http://mirrorlist.centos.org/?release=6&arch=$basearch&repo=os
gpgcheck=1
gpgkey=http://mirror.centos.org/centos/RPM-GPG-KEY-CentOS-6
#released updates
[updates]
name=CentOS-6 -Updates - 163.com
baseurl=http://mirrors.163.com/centos/6/updates/$basearch/
#mirrorlist=http://mirrorlist.centos.org/?release=6&arch=$basearch&repo=updates
gpgcheck=1
gpgkey=http://mirror.centos.org/centos/RPM-GPG-KEY-CentOS-6
#additionalpackages that may be useful
[extras]
name=CentOS-6 -Extras - 163.com
baseurl=http://mirrors.163.com/centos/6/extras/$basearch/
#mirrorlist=http://mirrorlist.centos.org/?release=6&arch=$basearch&repo=extras
gpgcheck=1
gpgkey=http://mirror.centos.org/centos/RPM-GPG-KEY-CentOS-6
#additionalpackages that extend functionality of existing packages
[centosplus]
name=CentOS-6 -Plus - 163.com
baseurl=http://mirrors.163.com/centos/6/centosplus/$basearch/
#mirrorlist=http://mirrorlist.centos.org/?release=6&arch=$basearch&repo=centosplus
gpgcheck=1
enabled=0
gpgkey=http://mirror.centos.org/centos/RPM-GPG-KEY-CentOS-6
#contrib -packages by Centos Users
[contrib]
name=CentOS-6 -Contrib - 163.com
baseurl=http://mirrors.163.com/centos/6/contrib/$basearch/
#mirrorlist=http://mirrorlist.centos.org/?release=6&arch=$basearch&repo=contrib
gpgcheck=1
enabled=0
gpgkey=http://mirror.centos.org/centos/RPM-GPG-KEY-CentOS-6
[k2][ceph-extras-source]
name=Ceph Extras Sources
baseurl=http://ceph.com/packages/ceph-extras/rpm/centos6/SRPMS
enabled=1
priority=2
gpgcheck=1
type=rpm-md
gpgkey=https://ceph.com/git/?p=ceph.git;a=blob_plain;f=keys/release.asc
[k3]
[k4][ceph]
name=Ceph packagesfor $basearch
baseurl=http://ceph.com/rpm/rhel6/$basearch
enabled=1
gpgcheck=1
type=rpm-md
gpgkey=https://ceph.com/git/?p=ceph.git;a=blob_plain;f=keys/release.asc
[ceph-noarch]
name=Ceph noarchpackages
baseurl=http://ceph.com/rpm/rhel6/noarch
enabled=1
gpgcheck=1
type=rpm-md
gpgkey=https://ceph.com/git/?p=ceph.git;a=blob_plain;f=keys/release.asc
[ceph-source]
name=Ceph sourcepackages
baseurl=http://ceph.com/rpm/rhel6/SRPMS
enabled=0
gpgcheck=1
type=rpm-md
gpgkey=https://ceph.com/git/?p=ceph.git;a=blob_plain;f=keys/release.asc
[k5]
[k6][apache2-ceph-noarch]
name=Apache noarchpackages for Ceph
baseurl=http://gitbuilder.ceph.com/apache2-rpm-centos6-x86_64-basic/ref/master/
enabled=1
priority=2
gpgcheck=1
type=rpm-md
gpgkey=https://ceph.com/git/?p=ceph.git;a=blob_plain;f=keys/autobuild.asc
[apache2-ceph-source]
name=Apache sourcepackages for Ceph
baseurl=http://gitbuilder.ceph.com/apache2-rpm-centos6-x86_64-basic/ref/master/
enabled=0
priority=2
gpgcheck=1
type=rpm-md
gpgkey=https://ceph.com/git/?p=ceph.git;a=blob_plain;f=keys/autobuild.asc
[k7][fastcgi-ceph-basearch]
name=FastCGIbasearch packages for Ceph
baseurl=http://gitbuilder.ceph.com/mod_fastcgi-rpm-centos6-x86_64-basic/ref/master
enabled=1
priority=2yum
gpgcheck=1
type=rpm-md
gpgkey=https://ceph.com/git/?p=ceph.git;a=blob_plain;f=keys/autobuild.asc
[fastcgi-ceph-noarch]
name=FastCGInoarch packages for Ceph
baseurl=http://gitbuilder.ceph.com/mod_fastcgi-rpm-centos6-x86_64-basic/ref/master
enabled=1
priority=2
gpgcheck=1
type=rpm-md
gpgkey=https://ceph.com/git/?p=ceph.git;a=blob_plain;f=keys/autobuild.asc
[fastcgi-ceph-source]
name=FastCGI sourcepackages for Ceph
baseurl=http://gitbuilder.ceph.com/mod_fastcgi-rpm-centos6-x86_64-basic/ref/master
enabled=0
priority=2
gpgcheck=1
type=rpm-md
gpgkey=https://ceph.com/git/?p=ceph.git;a=blob_plain;f=keys/autobuild.asc
[k8][ceph-extras-source]
name=Ceph ExtrasSources
baseurl=http://ceph.com/packages/ceph-extras/rpm/centos6/SRPMS
enabled=1
priority=2
gpgcheck=1
type=rpm-md
gpgkey=https://ceph.com/git/?p=ceph.git;a=blob_plain;f=keys/release.asc
四:安装ceph软件包
4.1、安装ceph部署机 ,使用ceph-deploy方式安装部署
yum install -yceph-deploy
4.2、安装ceph存储集群
(1)、安装ceph必备软件 ==all
yum install -y snappy leveldb gdisk python-argparse gperftools-libs
(2)、安装ceph软件 ==all
yum install -y ceph
4.3、安装ceph对象网关 ==all
yum install -y yum-plugin-priorities
yum update -y
4.4、安装apachefastcgi ==all
(1)、安装apachefastcgi 需要yum install httpd mod_fastcgi,安装之前先执行下面的操作
yuminstall -y httpd mod_fastcgi
4.4.1、确认 FastCGImodule启用.
cat/etc/httpd/conf.d/fastcgi.conf|grep mod_fastcgi.so
输出结果如下:
LoadModulefastcgi_module modules/mod_fastcgi.so
4.4.2、确认 Rewritemodule 启用
cat/etc/httpd/conf/httpd.conf|grep mod_rewrite.so
输出结果如下:
LoadModulerewrite_module modules/mod_rewrite.so
4.4.3、编辑配置文件httpd.conf ,修改httpd.conf ServerName为本机主机名
vim/etc/httpd/conf/httpd.conf
echo"ServerName $HOSTNAME">> /etc/httpd/conf/httpd.conf
4.4.4、启动httpd进程
/etc/init.d/httpd restart
4.5、安装SSL (安装此步骤有报错)
yum install -ymod_ssl openssl
4.5.1、创建证书
生成私钥.
openssl genrsa-out ca.key 2048
4.5.2、生成CSR文件
openssl req -new-key ca.key -out ca.csr
ountry Name (2 letter code) |
使用国际标准组织(ISO)国码格式,填写2个字母的国家代号。中国请填写CN。 |
State or Province Name (full name) |
省份,比如填写Shanghai |
Locality Name (eg, city) |
城市,比如填写Shanghai |
Organization Name (eg, company) |
组织单位,比如填写公司名称的拼音 |
Organizational Unit Name (eg, section) |
比如填写IT Dept |
Common Name (eg, your websites domain name): |
行使 SSL 加密的网站地址。请注意这里并不是单指您的域名,而是直接使用 SSL 的网站名称 例如:pay.abc.com。 一个网站这里定义是: |
Email Address |
邮件地址,可以不填 |
A challenge password |
可以不填 |
An optional company name |
可以不填 |
4.5.3、生成证书
opensslx509 -req -days 365 -in ca.csr -signkey ca.key -out ca.crt
4.5.4、copy生成证书至apache监听位置.
cp ca.crt/etc/pki/tls/certs
cp ca.key/etc/pki/tls/private/ca.key
cp ca.csr/etc/pki/tls/private/ca.csr
4.5.5、修改apache ssl配置文件 /etc/httpd/conf.d/ssl.conf.
vi +105/etc/httpd/conf.d/ssl.conf
将
105SSLCertificateFile /etc/pki/tls/certs/localhost.crt
改为:
105 SSLCertificateFile/etc/pki/tls/certs/ca.crt
将
112 SSLCertificateKeyFile/etc/pki/tls/private/localhost.key
改为
112 SSLCertificateKeyFile/etc/pki/tls/private/ca.key
4.5.6:将证书文件传到其它主机
--delete 删除那些DST中SRC没有的文件
for ip in $(cat/workspace/cephlist.txt);do echo -----$ip-----;rsync -avp --delete /etc/pki/tls/certs$ip:/etc/pki/tls/;done
for ip in $(cat/workspace/cephlist.txt);do echo -----$ip-----;rsync -avp --delete /etc/pki/tls/private$ip:/etc/pki/tls/;done
for ip in $(cat/workspace/cephlist.txt);do echo -----$ip-----;rsync -avp --delete/etc/httpd/conf.d/ssl.conf $ip:/etc/httpd/conf.d/ssl.conf;done
4.5.7、批量重启服务
for ip in $(cat /workspace/cephlist.txt);do echo-----$ip-----;ssh root@$ip /etc/init.d/httpd restart;done
官网步骤,写了host文件略过啦
(10)、AddWildcard to DNS (The address of the DNS must also be specified in the Cephconfiguration file with the rgw dns name = {hostname} setting.
address=/.ceph-node/192.168.0.1
五:安装ceph对象网关
5.1、基础软件包
yum install -y ceph-radosgwceph
5.2、安装ceph对象网关代理
yum install -yradosgw-agent
5.3、为块存储安装虚拟化软件
5.3.1、如果之前有qemu模块先删除,确保之后安装的是最完善的
yumremove -y qemu-kvm qemu-kvm-tools qemu-img
5.3.2、卸载后重新安装qemu
yum install -yqemu-kvm qemu-kvm-tools qemu-img
5.4、安装qemu客户代理
yum install -yqemu-guest-agent
5.5、在所有节点安装ceph依赖的软件及模块
yum install *argparse* -y
yum install redhat-lsb -y
yum install xfs* -y
############官方文档无此内容############
(4)、安装libvirt软件包
yuminstall -y libvirt
以上步骤参考配置连接
http://docs.ceph.com/docs/master/install/install-ceph-gateway/
配置 ceph对象网关
http://docs.ceph.com/docs/master/radosgw/config/
六、搭建ceph集群
http://docs.ceph.com/docs/master/radosgw/config/
6.1、配置mon节点
6.1.1、建立第一个mon节点
1、登录监控节点node01节点
ls /etc/ceph #查看ceph配置文件目录是否存在
2、创建ceph配置文件并配置ceph配置文件内的内容
3、创建一个ceph配置文件
touch/etc/ceph/ceph.conf
4、执行此命令可以得到一个唯一的标识,作为ceph集群ID
uuidgen
bc5ec6c4-3cc9-48ad-94d9-cee3cd75952a
fsid = bc5ec6c4-3cc9-48ad-94d9-cee3cd75952a
#此标识就是上面得到的,把此条命令加入ceph的配置文件
mon initialmembers = node01,node02,node03
#node01、node02、node03作为ceph集群的监控节点,把此条命令加入到ceph配置文件
mon host = 172.16.2.27,172.16.2.28, 172.16.2.29
#监控节点的地址,把此条命令加入ceph的配置文件中
5、ceph配置文件预览
vi /etc/ceph/ceph.conf
[global]
fsid = bc5ec6c4-3cc9-48ad-94d9-cee3cd75952a
mon initial members = node01,node02,node03
mon host = 172.16.2.27, 172.16.2.28, 172.16.2.29
public network = 172.16.2.0/24
auth cluster required = cephx
auth service required = cephx
auth client required = cephx
osd journal size = 1024
filestore xattr use omap = true
osd pool default size = 3
osd pool default min size = 1
osd crush chooseleaf type = 1
osd_mkfs_type = xfs
max mds = 5
mds max file size = 100000000000000
mds cache size = 1000000
mon osd down out interval = 900
#设置osd节点down后900s,把此osd节点逐出ceph集群,把之前映射到此节点的数据映射到其他节点。
cluster_network = 172.16.2.0/24
[mon]
mon clock drift allowed = .50
#把时钟偏移设置成0.5s,默认是0.05s,由于ceph集群中存在异构PC,导致时钟偏移总是大于0.05s,为了方便同步直接把时钟偏移设置成0.5s
6、在node01创建各种密钥
#为监控节点创建管理密钥
ceph-authtool--create-keyring /tmp/ceph.mon.keyring --gen-key -n mon. --cap mon 'allow *'
#为ceph amin用户创建管理集群的密钥并赋予访问权限
ceph-authtool--create-keyring /etc/ceph/ceph.client.admin.keyring --gen-key -n client.admin--set-uid=0 --cap mon 'allow *' --cap osd 'allow *' --cap mds 'allow'
#添加client.adminkey到 ceph.mon.keyring
ceph-authtool/tmp/ceph.mon.keyring --import-keyring /etc/ceph/ceph.client.admin.keyring
验证
cat/tmp/ceph.mon.keyring
7、在node01监控节点创建一个mon数据目录
mkdir -p/var/lib/ceph/mon/ceph-node01
8、在node01创建一个boot引导启动osd的key
mkdir -p/var/lib/ceph/bootstrap-osd/
ceph-authtool -C /var/lib/ceph/bootstrap-osd/ceph.keyring
9、在node01节点上初始化mon节点,
ceph-mon--mkfs -i node01 --keyring /tmp/ceph.mon.keyring
10、为了防止重新被安装创建一个空的done文件
touch/var/lib/ceph/mon/ceph-node01/done
11、创建一个空的初始化文件
touch/var/lib/ceph/mon/ceph-node01/sysvinit
12、启动ceph进程
/sbin/serviceceph -c /etc/ceph/ceph.conf start mon.node01
10、查看asok mon状态
[root@node01 ~]# ceph--cluster=ceph --admin-daemon /var/run/ceph/ceph-mon.node01.asok mon_status
6.1.2、添加其他mon(2)节点:
1、复制node01节点的/etc/ceph目录到node02
2、复制node01节点上的/var/lib/ceph/bootstrap-osd/ceph.keyring文件到node02
3、复制node01节点上的/tmp/ceph.mon.keyring
4、在node02节点上新建一个/var/lib/ceph/bootstrap-osd/目录
5、在node02节点上建立一个/var/lib/ceph/mon/ceph-node02目录
6、在node02节点上初始化mon节点
7、为了防止重新被安装创建一个空的done文件
8、创建一个空的初始化文件
9、启动ceph进程
scp /etc/ceph/*node02:/etc/ceph/
scp /var/lib/ceph/bootstrap-osd/ceph.keyringnode02:/var/lib/ceph/bootstrap-osd/
scp /tmp/ceph.mon.keyring node02:/tmp/
mkdir /var/lib/ceph/bootstrap-osd/
mkdir -p /var/lib/ceph/mon/ceph-node02
ceph-mon--mkfs -i node02 --keyring /tmp/ceph.mon.keyring
touch/var/lib/ceph/mon/ceph-node02/done
touch/var/lib/ceph/mon/ceph-node02/sysvinit
/sbin/service ceph -c/etc/ceph/ceph.conf start mon.node02
验证状态:
ceph --cluster=ceph--admin-daemon /var/run/ceph/ceph-mon.node02.asok mon_status
6.1.3、添加其他mon(3)节点:
添加方法同 6.1.2 步骤,注意修改node名称
6.2:添加osd节点
6.2.1、添加第一块osd节点
1、创建一个OSD,生成一个osdnumber
[root@node01 ~]# ceph osd create
0
2、为osd节点创建一个osd目录
[root@node01~]# mkdir -p/var/lib/ceph/osd/ceph-0
3、格式化已准备好的osd硬盘(格式化为xfs格式)
[root@node01 ~]# mkfs.xfs-f /dev/sdb
4、创建osd挂载目录
[root@node01 ~]#mount /dev/sdb /var/lib/ceph/osd/ceph-0
[root@node01 ~]#mount -o remount,user_xattr /var/lib/ceph/osd/ceph-0
查看挂载的情况
[root@node01 ~]#mount
/dev/sda2 on /type ext4 (rw)
proc on /proc typeproc (rw)
/dev/sdbon /var/lib/ceph/osd/ceph-1 type xfs (rw,user_xattr)
5、把上面的挂载信息写入分区表
[root@node01 ~]# vi /etc/fstab
/dev/sdb /var/lib/ceph/osd/ceph-0 xfs defaults 0 0
/dev/sdb /var/lib/ceph/osd/ceph-0 xfs remount,user_xattr 0 0
6、初始化osd数据目录
[root@node01 ~]# ceph-osd-i 0 --mkfs --mkkey
2015-04-0720:43:08.092553 7fab822b2800 -1 journal FileJournal::_open: disabling aio fornon-block journal. Use journal_force_aioto force use of aio anyway
2015-04-0720:43:08.181608 7fab822b2800 -1 journal FileJournal::_open: disabling aio fornon-block journal. Use journal_force_aioto force use of aio anyway
2015-04-0720:43:08.202946 7fab822b2800 -1 filestore(/var/lib/ceph/osd/ceph-0) could notfind 23c2fcde/osd_superblock/0//-1 in index: (2) No such file or directory
2015-04-0720:43:08.236009 7fab822b2800 -1 created object store /var/lib/ceph/osd/ceph-0journal /var/lib/ceph/osd/ceph-0/journal for osd.0 fsidfe3655d0-80d9-4dcf-a2c8-dcd1caf819ff
2015-04-0720:43:08.236076 7fab822b2800 -1 auth: error reading file:/var/lib/ceph/osd/ceph-0/keyring: can't open /var/lib/ceph/osd/ceph-0/keyring:(2) No such file or directory
2015-04-0720:43:08.236182 7fab822b2800 -1 created new key in keyring/var/lib/ceph/osd/ceph-0/keyring
6、注册osd的认证密钥
ceph auth addosd.[osd ID] osd 'allow *' mon 'allow profile osd' -i/var/lib/ceph/osd/ceph-[osd ID]/keyring
[root@node01 ~]#ceph auth add osd.0osd 'allow *' mon 'allow profile osd' -i /var/lib/ceph/osd/ceph-0/keyring
查看认证节点
ceph auth list
osd.1
key:AQDwSsZUQMkeGRAAmYYw9CoUUdK73IjDSH/gnw==
caps: [mon] allow profile osd
7、为此osd节点创建一个crushmap
[root@node01 ~]#ceph osd crush add-bucket node01 host
8、防放置 Node01 到根路径
[root@node01 ~]#ceph osd crush move node01 root=default
moved item id -2name 'node01' to location {root=default} in crush map
9、放置osd.0到 bucketnode01
[root@node01 ~]#ceph osd crush add osd.0 1.0 host=node01
add item id 0 name'osd.0' weight 1 at location {host=node01} to crush map
10、创建一个初始化目录
[root@node01 ~]#touch /var/lib/ceph/osd/ceph-0/sysvinit
11、启动osd进程
/etc/init.d/cephstart osd.0
12、查看osd目录树
[root@node01 ~]#ceph osd tree
# id weight type name up/down reweight
-1 1 root default
-2 1 host node01
0 1 osd.0 up 1
6.2.2、添加第二个osd节点
[root@node02 ~]#
1、创建一个OSD,生成一个osd number
2、为osd节点创建一个osd目录
3、格式化已准备好的osd硬盘,并挂在到上一步创建的osd目录(格式化为xfs格式)
4、挂在目录
5、初始化osd数据目录
6、注册osd的认证密钥
7、为此osd节点创建一个crush map
8、Place the Ceph Node under the root default
9、放置osd.0到 bucket node01
10、创建一个初始化目录
11、启动osd进程 、对应参数如下
ceph osd create
mkdir -p/var/lib/ceph/osd/ceph-1
mkfs.xfs-f /dev/sdb
mount/dev/sdb /var/lib/ceph/osd/ceph-1
mount -oremount,user_xattr /var/lib/ceph/osd/ceph-1
ceph-osd-i 1 --mkfs --mkkey
cephauth add osd.1 osd 'allow *' mon 'allow profile osd' -i /var/lib/ceph/osd/ceph-1/keyring
cephosd crush add-bucket node02 host
cephosd crush move node02 root=default
cephosd crush add osd.1 1.0 host=node02
touch/var/lib/ceph/osd/ceph-1/sysvinit
/etc/init.d/cephstart osd.1
12、把上面的挂载信息写入分区表
vi /etc/fstab
/dev/sdb /var/lib/ceph/osd/ceph-1 xfs defaults 0 0
/dev/sdb /var/lib/ceph/osd/ceph-1 xfs remount,user_xattr 0 0
验证:
mount
/dev/sdb on/var/lib/ceph/osd/ceph-1 type xfs (rw,user_xattr)
13、查看osd目录树
[root@node02 ~]#ceph osd tree
# id weight type name up/down reweight
-1 2 root default
-2 1 host node01
0 1 osd.0 up 1
-3 1 host node02
1 1 osd.1 up 1
6.2.3、添加其它osd节点 ,参考6.2.1、6.2.2 步骤
添加node01 服务器上的其它磁盘为osd[k1]
添加node02 服务器上的其它磁盘为osd[k2] [k3]
添加node03 服务器上的其它磁盘为osd[k4] [k5]
验证:
mount
12、查看osd目录树
[root@node02 ~]#ceph osd tree
添加其它osd方法步骤同上
添加删除osd参考文档
http://docs.openfans.org/ceph/ceph4e2d658765876863/ceph-1/ceph-storage-cluster3010ceph5b5850a896c67fa43011/operations301064cd4f5c3011/adding-removing-osds3010589e52a0-52209664osd3011
6.3:添加元数据服务器
6.3.1、添加第一个元数据服务器
1、为mds元数据服务器创建一个目录
[root@node1 ~]#mkdir -p /var/lib/ceph/mds/ceph-node01
2、为bootstrap-mds客户端创建一个密钥 注:(如果下面的密钥在目录里已生成可以省略此步骤)
[root@node1 ~]#ceph-authtool --create-keyring /var/lib/ceph/bootstrap-mds/ceph.keyring --gen-key-n client.bootstrap-mds
3、在ceph auth库中创建bootstrap-mds客户端,赋予权限添加之前创建的密钥
注(查看ceph auth list 用户权限认证列表 如果已有client.bootstrap-mds此用户,此步骤可以省略)
[root@node1 ~]#ceph auth add client.bootstrap-mds mon 'allow profile bootstrap-mds' -i/var/lib/ceph/bootstrap-mds/ceph.keyring
added key forclient.bootstrap-mds
4、在root家目录里创建ceph.bootstrap-mds.keyring文件
touch/root/ceph.bootstrap-mds.keyring
5、把keyring/var/lib/ceph/bootstrap-mds/ceph.keyring里的密钥导入家目录下的ceph.bootstrap-mds.keyring文件里
ceph-authtool--import-keyring /var/lib/ceph/bootstrap-mds/ceph.keyring ceph.bootstrap-mds.keyring
6、在ceph auth库中创建mds.node01用户,并赋予权限和创建密钥,密钥保存在/var/lib/ceph/mds/ceph-node01/keyring文件里
ceph--cluster ceph --name client.bootstrap-mds --keyring/var/lib/ceph/bootstrap-mds/ceph.keyring auth get-or-create mds.node01 osd'allow rwx' mds 'allow' mon 'allow profile mds' -o /var/lib/ceph/mds/ceph-node01/keyring
7、为mds创建一个初始化文件用于启动使用(此文件为空文件)
[root@node1 ~]# touch/var/lib/ceph/mds/ceph-node01/sysvinit
8、为了防止重新被安装创建一个空的done文件
[root@node1 ~]#touch /var/lib/ceph/mds/ceph-node01/done
9、情况mds服务进程
[root@node1 ~]#service ceph start mds.node01
=== mds.node01 ===
Starting Ceph mds.node01on node1...
starting mds.node01at :/0
6.3.2:添加其它元数据服务器,参考6.3.1 步骤
添加第二个元数据服务器[k6]
添加第三个元数据服务器[k7]
1、查看osd目录树
ceph osd tree
查看集群状态
ceph -w
七:安装client端RBD、cephFS挂载演示
7.1:客户端内核要求、安装装准备
7.1.1、升级client端内核
A. 操作系统使用CentOS6.6,因为系统已经包含xfs的支持可以直接使用不需要再次编译。
B. 由于CentOS6.5系统的内核为2.6.32,因此要关闭硬盘的写入缓存,若高于此版本不需要关闭。
i. #hdparm -W 0/dev/sdb 0
7.1.2:升级系统内核
kernel 2.6.34以前的版本是没有Modulerbd的,把系统内核版本升级到最新
rpm --import http://elrepo.org/RPM-GPG-KEY-elrepo.org
rpm -Uvhhttp://elrepo.org/elrepo-release-6-5.el6.elrepo.noarch.rpm
yum--enablerepo=elrepo-kernel install kernel-ml -y
安装完内核后修改/etc/grub.conf配置文件使
修改配置文件中的 Default=1 toDefault=0
7.1.3、验证内核支持
[首先加载ceph.ko]
#modprobe -l|grep ceph
kernel/fs/ceph/ceph.ko
kernel/net/ceph/libceph.ko
#modprobe ceph
机器重启后生效 init 6
7.2、RBD方式挂载
7.2.1、安装ceph软件包
Yum install ceph –y
7.2.2、将node01配置文件copyclient
rsync -avp node01:/etc/ceph/* /etc/ceph/
7.2.3、新建一个cephpool
ceph osd pool create{pool-name} {pg-num} [{pgp-num}]
[root@client ~]# cephosd pool create kman 256
7.2.3、在pool中新建一个镜像
[root@client ~]# rbd create rbdtest-1--size 80960 -p kman
7.2.4、把镜像映射到pool块设备中
[root@client ~]# rbd map rbdtest-1-p kman
取消映射块设备:
使用RBD命令要取消映射块设备镜像,指定rm选项,设备名称(即,一般以作为块设备映像名称相同)
sudo rbd unmap/dev/rbd/{poolname}/{p_w_picpathname}
例如:
sudo rbd unmap/dev/rbd/rbd/foo
7.2.5、查看镜像映射map
[root@client ~]# rbdshowmapped
id pool p_w_picpath snap device
0 kman rbdtest-1 - /dev/rbd0
7.2.6、格式化映射的设备块
[root@client ~]# mkfs.ext4 -m0/dev/rbd0
参数:-m reserved-blocks-percentage
保留块百分比,-m:设置预留空间
默认是5%。也就是2G的空间,文件系统只是有2G*95%
7.2.7、挂载新建的分区
[root@client ~]# mkdir /rbdtest
[root@client ~]# mount/dev/rbd0 /rbdtest
[root@client ~]# df -h
/dev/sda5 1.7T 4.8G 1.7T 1% /data
/dev/rbd0 78G 56M 78G 1% /rbdtest
7.2.8、进入新建的分区并dd测试性能
[root@client ~]# cd /rbdtest
7.2.9、其它参考命令
查看创建的池
ceph osd lspools
删除池
ceph osd pool delete kman kman--yes-i-really-really-mean-it
创建池参考资料
http://docs.openfans.org/ceph/ceph4e2d658765876863/ceph-1/ceph-storage-cluster3010ceph5b5850a896c67fa43011/operations301064cd4f5c3011/pools30106c603011
7.3、cephFS挂载
参考文档
http://docs.openfans.org/ceph/ceph4e2d658765876863/ceph-1/copy_of_ceph-filesystem301065874ef67cfb7edf3011/manpage-mount-ceph3010mount-ceph5e2e52a998753011
创建cephfs文件系统
对于一个刚创建的MDS服务,虽然服务是运行的,但是它的状态直到创建 pools 以及文件系统的时候才会变为Active.
[root@zeus-4-17 rbdtest]# ceph mds dump
dumped mdsmap epoch 1
epoch 1
flags 0
created 0.000000
modified 2015-01-26 21:49:35.368742
tableserver 0
root 0
session_timeout 0
session_autoclose 0
max_file_size 0
last_failure 0
last_failure_osd_epoch 0
compat compat={},rocompat={},incompat={}
max_mds 0
in
up {}
failed
stopped
data_pools
metadata_pool 0
inline_data disabled
7.3.1、创建cephfs 存储池
ceph osd pool create {pool-name} {pg-num}[{pgp-num}]
{
Option implication:
{pool-name}--> 存储池名称,必须唯一。
{pg-num}--> 存储池拥有的归置组总数。
{pgp-num}--> 用于归置的归置组总数。
ceph osd pool create cephfs_data10
ceph osd pool createcephfs_metadata 10
ceph fs newleadorfs cephfs_metadata cephfs_data
验证数据生成
[root@zeus-4-17 rbdtest]# ceph mds dump
dumped mdsmap epoch 5
epoch 5
flags 0
created 2015-01-27 16:10:16.789446
modified 2015-01-27 16:10:21.504363
tableserver 0
root 0
session_timeout 60
session_autoclose 300
max_file_size 100000000000000
last_failure 0
last_failure_osd_epoch 0
compat compat={},rocompat={},incompat={1=base v0.20,2=client writeableranges,3=default file layouts on dirs,4=dir inode in separate object,5=mds usesversioned encoding,6=dirfrag is stored in omap,8=no anchor table}
max_mds 5
in 0
up {0=4341}
failed
stopped
data_pools 2
metadata_pool 3
inline_data disabled
4341: 172.16.2.29:6815/31531 'node03' mds.0.1 up:active seq 1493
7.3.2、客户端挂载
1、安装客户端挂载软件
yum install ceph-fuse -y
2、创建挂载目录
mkdir /mycephfs
3、模拟挂载
mount -t ceph node01:6789:/ /mycephfs -v -oname=admin,secret=AQCnrMZUaFsiCxAAXzM3aF9WjUBnwbN6PtvZEw==
返回结果:
parsing options:rw,name=admin,secret=AQCnrMZUaFsiCxAAXzM3aF9WjUBnwbN6PtvZEw==
4、证书来源位置
more /etc/ceph/ceph.client.admin.keyring #key证书来源
[client.admin]
key = AQCnrMZUaFsiCxAAXzM3aF9WjUBnwbN6PtvZEw==
auid = 0
caps mds = "allow"
caps mon = "allow *"
caps osd = "allow *"
5、另外一种命令挂载方式
[root@ceph-client ~]# mount -t ceph node01:6789:/ /mycephfs -v -oname=admin,secretfile=/etc/ceph/ceph.client.admin.keyring
6、若果有多个mon监控节点,可以挂载多可节点,保证了cephFS的安全行,当有一个节点down的时候不影响写入数据
[root@client ~]# mount.cephnode01,node02,node03:/ /mycephfs -v -o name=admin,secret= AQCnrMZUaFsiCxAAXzM3aF9WjUBnwbN6PtvZEw==
7、验证挂载信息:
df -h
Filesystem Size Used Avail Use% Mounted on
/dev/sda5 1.7T 4.8G 1.7T 1% /data
/dev/rbd0 78G 56M 78G 1% /rbdtest
172.16.2.27:6789:/ 8.2T 9.8G 8.2T 1% /mycephfs
8、把挂载的信息写到fstab里
[root@client ~]# vi /etc/fstab
172.16.2.27,172.16.2.28,172.16.2.29:/ /mycephfs ceph name=admin,secret= AQCnrMZUaFsiCxAAXzM3aF9WjUBnwbN6PtvZEw==,noatime 0
9、指定key文件的调用方式
sudoceph-fuse -k ./ceph.client.admin.keyring -m 192.168.40.107:6789 ~/mycephfs
10、挂载后查看结果,注意观察类型
df-Th
11、取消挂载
取消挂载的操作如下:
sudoumount /mnt/mycephfs
八:检查ceph集群状态常用命令
#ceph health //ceph健康状态
#ceph status //ceph当前全部状态
#ceph -w //实时监控ceph状态及变化
#ceph osddump //所有osd详细状态
#ceph osd tree //osd所在位置,及状态
#cephquorum_status //mon优先级状态
#ceph mon dump //mon节点状态
#ceph mds dump //mds详细状态
8.1、磁盘压力测试软件推荐
安装fio压测工具、Linux系统自带,yum即可安装
yum install fio –y
http://www.360doc.com/content/12/0809/11/2459_229177650.shtml
http://www.gongzi.org/linux-io-test.html
九:故障排查
9.1、创建osd节点挂载磁盘报错,报错信息如下
[root@node01 ~]#mount -o user_xattr /dev/sdb /var/lib/ceph/osd/ceph-0
mount:wrong fs type, bad option, bad superblock on /dev/sdb,
missing codepage or helper program, orother error
In some cases useful info is found in syslog -try
dmesg | tail or so
执行上面的命令会报错
解决的办法是:用下面的两条命令替代上面的一条命令。
[root@node01~]# mount /dev/sdb /var/lib/ceph/osd/ceph-0
[root@node01~]# mount -o remount,user_xattr /var/lib/ceph/osd/ceph-0
9.2:执行ceph 命令报错
Traceback (mostrecent call last):
File "/usr/bin/ceph", line 63, in
import rados
ImportError: Nomodule named rados
解决:注释自己安装python的环境变量
[root@vhadoop-18-14~]# more /etc/profile
#exportPYTHON=/opt/python/
#exportPATH=$PYTHON/bin:$PATH
#其它类似报错
Traceback (mostrecent call last):
File"/usr/local/bin/ceph", line 53, in
import argparse
ImportError: Nomodule named argparse
解决办法:
yuminstall python-pip
pipinstall argparse
pipinstall argparse --upgrade
/usr/local/bin/cephosd create
Traceback (mostrecent call last):
File"/usr/local/bin/ceph", line 56, in
import rados
ImportError: Nomodule named rados cp -vf /usr/local/ceph/lib/python2.6/site-packages/*/usr/lib64/python2.6
echo/usr/local/ceph/lib >/etc/ld.so.conf.d/ceph.conf
ldconfig
9.3:创建osd报错
[root@node01 ~]#ceph-osd -i 3 --mkfs --mkkey
2015-01-2712:29:06.775251 7f4c5d342800 -1 OSD::mkfs: ObjectStore::mkfs failed with error-16
2015-01-2712:29:06.775288 7f4c5d342800 -1 **ERROR: error creating empty object store in /var/lib/ceph/osd/ceph-3: (16)Device or resource busy
排查解决:
[root@node01 ~]#mount 查看挂载分区
/dev/sda2 on /type ext4 (rw)
proc on /proc typeproc (rw)
sysfs on /sys typesysfs (rw)
devpts on /dev/ptstype devpts (rw,gid=5,mode=620)
tmpfs on /dev/shmtype tmpfs (rw)
/dev/sda1 on /boottype ext4 (rw)
none on/proc/sys/fs/binfmt_misc type binfmt_misc (rw)
/dev/sdb on/var/lib/ceph/osd/ceph-0 type xfs (rw,user_xattr)
此处的磁盘挂载出错啦,应该是一个新磁盘 /dev/sdc ,
/dev/sdbon /var/lib/ceph/osd/ceph-3 type xfs (rw,user_xattr)
9.4:报错
在ceph auth库中创建mds.node01用户,并赋予权限和创建密钥,密钥保存在/var/lib/ceph/mds/ceph-node01/keyring文件里
root@node01 ~]# ^C
[root@node01 ~]#ceph --cluster ceph --name client.bootstrap-mds --keyring/var/lib/ceph/bootstrap-mds/ceph.keyring auth get-or-create mds.node01 osd'allow rwx' mds 'allow' mon 'allow profile mds' -o/var/lib/ceph/mds/ceph-node01/keyring
2015-01-27 13:54:44.1895357f79f9ec5700 0 librados:client.bootstrap-mds authentication error (1) Operation not permitted
Error connectingto cluster: PermissionError
解决办法:
确认key文件信息:
more/var/lib/ceph/bootstrap-mds/ceph.keyring
[client.bootstrap-mds]
key = AQDvRcZUALD4MxAAv4s4cMr6XraBt+Pw2Iowyg==
通过ceph auth list 命令查看审核信息,找到
client.bootstrap-mds
key:AQDvRcZUALD4MxAAv4s4cMr6XraBt+Pw2Iowyg==
caps: [mon] allow profile bootstrap-mds
确保/var/lib/ceph/bootstrap-mds/ceph.keyring 文件内容与ceph auth list 查看信息一致
9.5:报错
ceph -w
clusterbc5ec6c4-3cc9-48ad-94d9-cee3cd75952a
health HEALTH_WARN too few pgs per osd (7 < min 20)
monmap e1: 3 mons at{node01=172.16.2.27:6789/0,node02=172.16.2.28:6789/0,node03=172.16.2.29:6789/0},election epoch 8, quorum 0,1,2 node01,node02,node03
osdmap e73: 9 osds: 9 up, 9 in
pgmap v176: 64 pgs, 1 pools, 0 bytesdata, 0 objects
9524 MB used, 8365 GB / 8374 GB avail
64 active+clean
解决办法:
解决办法, 需要修改pg_num , pgp_num .
先要获取pool name, 如下, 返回pool name : rbd.
ceph osd poolstats
pool rbd id 0
nothing is going on
修改pool对应的pg_num和pgp_num.
ceph osd pool set rbdpg_num 128
specified pg_num128 <= current 128 #要大于当前128
cephosd pool set rbd pg_num 300
验证:
ceph -s
clusterbc5ec6c4-3cc9-48ad-94d9-cee3cd75952a
health HEALTH_WARN too few pgs per osd (14< min 20); pool rbd pg_num 128 > pgp_num 64
monmap e1: 3 mons at{node01=172.16.2.27:6789/0,node02=172.16.2.28:6789/0,node03=172.16.2.29:6789/0},election epoch 8, quorum 0,1,2 node01,node02,node03
osdmap e75: 9 osds: 9 up, 9 in
pgmap v184: 128 pgs, 1 pools, 0 bytesdata, 0 objects
9527 MB used, 8365 GB / 8374 GB avail
128 active+clean
同时还需要调整pgp_num
[root@mon1~]# ceph osd pool set rbd pgp_num 300
ceph -s
clusterbc5ec6c4-3cc9-48ad-94d9-cee3cd75952a
health HEALTH_WARN toofew pgs per osd (14 < min 20)
monmap e1: 3 mons at{node01=172.16.2.27:6789/0,node02=172.16.2.28:6789/0,node03=172.16.2.29:6789/0},election epoch 8, quorum 0,1,2 node01,node02,node03
osdmap e78: 9 osds: 9 up, 9 in
pgmap v225: 128 pgs, 1 pools, 0 bytesdata, 0 objects
9529 MB used, 8365 GB / 8374 GBavail
128 active+clean
最终验证:
[root@node02 ~]#ceph -s
clusterbc5ec6c4-3cc9-48ad-94d9-cee3cd75952a
health HEALTH_OK
monmap e1: 3 mons at{node01=172.16.2.27:6789/0,node02=172.16.2.28:6789/0,node03=172.16.2.29:6789/0},election epoch 8, quorum 0,1,2 node01,node02,node03
osdmap e83: 9 osds: 9 up, 9 in
pgmap v265: 300 pgs, 1 pools, 0 bytesdata, 0 objects
9538 MB used, 8365 GB / 8374 GBavail
300 active+clean
9.6: 报错
[root@node01 ~]#ceph -w
clusterbc5ec6c4-3cc9-48ad-94d9-cee3cd75952a
health HEALTH_WARN pool cephfs_data has too fewpgs