GlusterFS是一个可扩展,分布式文件系统,集成来自多台服务器上的磁盘存储资源到单一全局命名空间,以提供共享文件存储。
特点:
GFS存储的一些术语:
Brick:GlusterFS中的存储单元,通常是一个受信存储池中的服务器的一个目录。可以通过主机名和目录名来标识,如'SERVER:EXPORT'。
Node:一个拥有若干brick的设备。
Volume:一组bricks的逻辑集合。
Client:挂载了GlusterFS卷的设备。
GFID:GlusterFS卷中的每个文件或目录都有一个唯一的128位的数据相关联,其用于模拟inode
Namespace:每个Gluster卷都导出单个ns作为POSIX的挂载点。
RDMA:远程直接内存访问,支持不通过双方的OS进行直接内存访问。
RRDNS:round robin DNS是一种通过DNS轮转返回不同的设备以进行负载均衡的方法
Self-heal:用于后台运行检测复本卷中文件和目录的不一致性并解决这些不一致。
Split-brain:脑裂
Volfile:Glusterfs进程的配置文件,通常位于/var/lib/glusterd/vols/volname
GFS卷(volume)的模式:
Volume是一组Brick的组合,一个gfs集群中可以有多个volume,可以供客户端挂载、存储数据等,它有很多种模式供选择:
主机名
ip地址
角色
master
192.168.72.100
k8s-master、glusterfs、heketi
node1
192.168.72.101
k8s-work节点
node2
192.168.72.102
k8s-work节点
获取gluster rpm包,我这里是7.5版本的,需要下载以下几个rpm包:
# ls
glusterfs-7.5-1.el7.x86_64.rpm glusterfs-libs-7.5-1.el7.x86_64.rpm
glusterfs-api-7.5-1.el7.x86_64.rpm glusterfs-server-7.5-1.el7.x86_64.rpm
glusterfs-api-devel-7.5-1.el7.x86_64.rpm psmisc-22.20-15.el7.x86_64.rpm
glusterfs-cli-7.5-1.el7.x86_64.rpm userspace-rcu-0.10.0-3.el7.x86_64.rpm
glusterfs-client-xlators-7.5-1.el7.x86_64.rpm userspace-rcu-0.7.16-1.el7.x86_64.rpm
glusterfs-fuse-7.5-1.el7.x86_64.rpm
地址:https://buildlogs.centos.org/centos/7/storage/x86_64/gluster-9/Packages/;如果有外网可以直接配置yum源安装需要的版本。
安装rpm包
## 先安装依赖的环境包
yum -y install attr psmisc rpcbind
## 安装glusterfs
rpm -ivh glusterfs-libs-7.5-1.el7.x86_64.rpm
rpm -ivh glusterfs-7.5-1.el7.x86_64.rpm
rpm -ivh glusterfs-api-7.5-1.el7.x86_64.rpm glusterfs-cli-7.5-1.el7.x86_64.rpm glusterfs-client-xlators-7.5-1.el7.x86_64.rpm
rpm -ivh glusterfs-fuse-7.5-1.el7.x86_64.rpm
rpm -ivh userspace-rcu-0.10.0-3.el7.x86_64.rpm
rpm -ivh glusterfs-server-7.5-1.el7.x86_64.rpm
rpm -ivh glusterfs-api-devel-7.5-1.el7.x86_64.rpm
因为包之间会有依赖关系,按照以上顺序可以顺利安装,如果不想这么麻烦,可强制安装--force --nodeps。
启动服务
## 需要启动两个服务glusterd和glusterfsd
systemctl start glusterd glusterfsd
glusterd和glusterfsd两个服务的区别,以及gluster和glusterfs命令
glusterd是其服务的守护进程,是一个管理模块,处理gluster发过来的命令,处理集群管理、存储池管理、brick管理、负载均衡、快照管理等,默认端口24007:
# ss -tnlp | grep 24007
LISTEN 0 128 *:24007 *:* users:(("glusterd",pid=7873,fd=10))
## 进程
# ps -ef | grep gluster
root 7873 1 0 Mar28 ? 00:00:41 /usr/sbin/glusterd -p /var/run/glusterd.pid --log-level INFO
glusterfsd是服务端模块,存储池中的每个brick(卷组)都会启动一个glusterfsd进程。此模块主要是处理客户端的读写请求,默认端口是从49152开始,后续端口都+1。我这里启动了两个卷组:
# ss -tnlp | grep gluster
LISTEN 0 128 *:49152 *:* users:(("glusterfsd",pid=7898,fd=11))
LISTEN 0 128 *:49153 *:* users:(("glusterfsd",pid=44275,fd=11))
## 进程
# ps -ef | grep gluster
root 7898 1 0 Mar28 ? 00:04:46 /usr/sbin/glusterfsd -s 192.168.72.100 --volfile-id xtest-fs.192.168.72.100.data-text-mind -p /var/run/gluster/vols/xtest-fs/192.168.72.100-data-text-mind.pid -S /var/run/gluster/3b5c03678c749dfd.socket --brick-name /data/text-mind -l /var/log/glusterfs/bricks/data-text-mind.log --xlator-option *-posix.glusterd-uuid=46bfd5ab-07be-4a97-993f-1ddab13e0ee9 --process-name brick --brick-port 49152 --xlator-option xtest-fs-server.listen-port=49152
glusterfs命令是一个客户端模块,负责通过mount挂载集群中某台服务器的存储池,以目录的形式呈现给用户。用户mount几个卷组,就是出现几个glusterfs的进程,如下:
# ps -ef | grep -w glusterfs | grep -v glusterfsd
root 31496 1 0 Mar25 ? 00:02:46 /usr/sbin/glusterfs --process-name fuse --volfile-server=yq01-aip-aikefu10.yq01.baidu.com --volfile-id=xtest-fs /data/mnt
gluster命令就是客户端请求命令,负责发起请求,比如创建、查看当前卷组等:
# gluster volume create xtest-fs 192.168.72.100:/data/xtest-fs/brick force
# gluster volume list
如果是多台gfs构成集群,在服务安装完毕之后,要进行一个添加的动作
## 添加
# gluster peer probe node1
# gluster peer probe node2
## 查看当前集群
# gluster pool list
heketi提供一个RESTful管理节点界面,可以用来管理GlusterFS卷的生命周期,通过heketi,就可以像使用Opentack Manila,kubernete和openShift一样申请可以动态配置GlusterFS卷,Heketi会动态在集群内选择bricks构建所需的volumes,这样以确保数据的副本会分散到集群不同的故障域内。
安装heketi服务,有源可以直接yum安装,没有可以单独下载rpm包安装,或者通过docker、k8s安装
# yum -y install heketi heketi-client
# heketi --version
Heketi 9.0.0
# systemctl start heketi.service
配置ssh秘钥,使heketi节点可以免密访问glusterfs节点
# ssh-keygen -f /etc/heketi/heketi_key -t rsa -N ''
# ssh-copy-id -i /etc/heketi/heketi_key.pub master
# chown heketi:heketi /etc/heketi/
编辑heketi的配置文件/etc/heketi/heketi.json
heketi有三种执行方式,分别为mock,ssh,kubernetes,官方建议在测试和开发环境使用mock,生产使用ssh,如果glusterfs是部署在k8s上,则使用kubernetes方式。
下边对配置部分做出解释:
{
## heketi服务的默认端口,可以更改。
"_port_comment": "Heketi Server Port Number",
"port": "8080",
## 是否开启认证,一般选是。
"_use_auth": "Enable JWT authorization. Please enable for deployment",
"use_auth": true,
## 若开启认证,则需要编辑认证的用户及密码。
"_jwt": "Private keys for access",
"jwt": {
"_admin": "Admin has access to all APIs",
"admin": {
"key": "My Secret"
},
"_user": "User only has access to /volumes endpoint",
"user": {
"key": "My Secret"
}
},
"_glusterfs_comment": "GlusterFS Configuration",
"glusterfs": {
"_executor_comment": [
"Execute plugin. Possible choices: mock, ssh",
"mock: This setting is used for testing and development.",
" It will not send commands to any node.",
"ssh: This setting will notify Heketi to ssh to the nodes.",
" It will need the values in sshexec to be configured.",
"kubernetes: Communicate with GlusterFS containers over",
" Kubernetes exec api."
],
## 选择操作gfs的方式,这里选择ssh。
"executor": "ssh",## 配置ssh秘钥
"_sshexec_comment": "SSH username and private key file information",
"sshexec": {
"keyfile": "/etc/heketi/heketi_key",
"user": "root",
"port": "22",
"fstab": "/etc/fstab"
},
## 配置k8s证书
"_kubeexec_comment": "Kubernetes configuration",
"kubeexec": {
"host" :"https://kubernetes.host:8443",
"cert" : "/path/to/crt.file",
"insecure": false,
"user": "kubernetes username",
"password": "password for kubernetes user",
"namespace": "OpenShift project or Kubernetes namespace",
"fstab": "Optional: Specify fstab file on node. Default is /etc/fstab"
},
"_db_comment": "Database file name",
## heketi的数据存储位置
"db": "/var/lib/heketi/heketi.db",
"_loglevel_comment": [
"Set log level. Choices are:",
" none, critical, error, warning, info, debug",
"Default is warning"
],
## heketi的日志级别
"loglevel" : "debug"
}
}
启动heketi服务,设置开启自启,测试访问
# systemctl start heketi
# systemctl enable heketi
# curl http://master:8080/hello
Hello from Heketi
配置heketi连接gfs服务
(1) 在集群的主节点设置环境变量,注意端口根据实际情况修改
# export HEKETI_CLI_SERVER=http://master:8080
(2) 在集群的主节点修改/usr/share/heketi/topology-sample.json配置文件,执行添加节点和添加device的操作:
{
"clusters": [
{
"nodes": [
{
"node": {
"hostnames": {
"manage": [
"master"
],
"storage": [
"192.168.72.100"
]
},
"zone": 1
},
"devices": [
{
"name": "/dev/sdb",
"destroydata": false
}
]
}
]
}
]
}
manage指定gfs主机域名,storage指定gfs主机ip,devices下指定要存储gfs数据的数据盘,此盘应该是一块未分区的干净盘,分区应该也可以,我这里没有试。
可以添加多个主机和多块盘,具体格式如下:
{
"clusters": [
{
"nodes": [
{
"node": {
"hostnames": {
"manage": [
"192.168.10.100"
],
"storage": [
"192.168.10.100"
]
},
"zone": 1
},
"devices": [
{
"name": "/dev/sdb",
"destroydata": false
},
{
"name": "/dev/sdc",
"destroydata": false
},
{
"name": "/dev/sdd",
"destroydata": false
},
{
"name": "/dev/sde",
"destroydata": false
},
{
"name": "/dev/sdf",
"destroydata": false
},
{
"name": "/dev/sdg",
"destroydata": false
},
{
"name": "/dev/sdh",
"destroydata": false
},
{
"name": "/dev/sdi",
"destroydata": false
}
]
},
{
"node": {
"hostnames": {
"manage": [
"192.168.10.101"
],
"storage": [
"192.168.10.101"
]
},
"zone": 2
},
"devices": [
{
"name": "/dev/sdb",
"destroydata": false
},
{
"name": "/dev/sdc",
"destroydata": false
},
{
"name": "/dev/sdd",
"destroydata": false
},
{
"name": "/dev/sde",
"destroydata": false
},
{
"name": "/dev/sdf",
"destroydata": false
},
{
"name": "/dev/sdg",
"destroydata": false
},
{
"name": "/dev/sdh",
"destroydata": false
},
{
"name": "/dev/sdi",
"destroydata": false
}
]
},
{
"node": {
"hostnames": {
"manage": [
"192.168.10.102"
],
"storage": [
"192.168.10.102"
]
},
"zone": 1
},
"devices": [
{
"name": "/dev/sdb",
"destroydata": false
},
{
"name": "/dev/sdc",
"destroydata": false
},
{
"name": "/dev/sdd",
"destroydata": false
},
{
"name": "/dev/sde",
"destroydata": false
},
{
"name": "/dev/sdf",
"destroydata": false
},
{
"name": "/dev/sdg",
"destroydata": false
},
{
"name": "/dev/sdh",
"destroydata": false
},
{
"name": "/dev/sdi",
"destroydata": false
}
]
},
{
"node": {
"hostnames": {
"manage": [
"192.168.10.103"
],
"storage": [
"192.168.10.103"
]
},
"zone": 2
},
"devices": [
{
"name": "/dev/sdb",
"destroydata": false
},
{
"name": "/dev/sdc",
"destroydata": false
},
{
"name": "/dev/sdd",
"destroydata": false
},
{
"name": "/dev/sde",
"destroydata": false
},
{
"name": "/dev/sdf",
"destroydata": false
},
{
"name": "/dev/sdg",
"destroydata": false
},
{
"name": "/dev/sdh",
"destroydata": false
},
{
"name": "/dev/sdi",
"destroydata": false
}
]
}
]
}
]
}
(3) 导入配置文件到heketi,使heketi管理gfs
# heketi-cli -s $HEKETI_CLI_SERVER --user admin --secret 'My Secret' topology load --json=/usr/share/heketi/topology-sample.json
Creating cluster ... ID: 6b7b8affdb10add55ea367e7e2f8e091
Allowing file volumes on cluster.
Allowing block volumes on cluster.
Creating node master ... ID: 96bbfbf66bc8ad551fe5a8c8ef289218
Adding device /dev/sdb ... OK
如果报错:Error: Unable to get topology information: Invalid JWT token: Token missing iss claim。是因为没加用户名和密码 --user admin --secret 'My Secret'。
(4) 测试创建一个volume,看是否成功
# heketi-cli volume create --size=10 --durability=none --user "admin" --secret "My Secret"
Name: vol_38412746b79cf0d7ffa69a3969c1cfed
Size: 10
Volume Id: 38412746b79cf0d7ffa69a3969c1cfed
Cluster Id: 6b7b8affdb10add55ea367e7e2f8e091
Mount: 192.168.72.100:vol_38412746b79cf0d7ffa69a3969c1cfed
Mount Options: backup-volfile-servers=
Block: false
Free Size: 0
Reserved Size: 0
Block Hosting Restriction: (none)
Block Volumes: []
Durability Type: none
Distribute Count: 1
创建成功,heketi已经可以管理gfs服务。
创建StorageClass
vi storageclass-gfs-heketi-distributed.yaml
apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
name: gluster-distributed
provisioner: kubernetes.io/glusterfs
reclaimPolicy: Retain
parameters:
resturl: "http://192.168.72.100:8080"
restauthenabled: "true"
restuser: "admin"
restuserkey: "My Secret"
gidMin: "40000"
gidMax: "50000"
volumetype: "none"
allowVolumeExpansion: true
volumeBindingMode: Immediate
provisioner:表示存储分配器,需要根据后端存储的不同而变更;
reclaimPolicy: 默认即“Delete”,删除pvc后,相应的pv及后端的volume,brick(lvm)等一起删除;设置为”Retain”时则保留数据,需要手工处理
resturl:heketi API服务提供的url;
restauthenabled:可选参数,默认值为"false",heketi服务开启认证时必须设置为"true";
restuser:可选参数,开启认证时设置相应用户名;
restuserkey:可选参数。开启认证时输入对应用户的密码;
secretNamespace:可选参数,开启认证时可以设置为使用持久化存储的namespace;
secretName:可选参数,开启认证时,需要将heketi服务的认证密码保存在secret资源中;
clusterid:可选参数,指定集群id,也可以是1个clusterid列表,格式为"id1,id2";
volumetype:可选参数,设置卷类型及其参数,如果未分配卷类型,则有分配器决定卷类型;如"volumetype: replicate:3"表示3副本的replicate卷,"volumetype: disperse:4:2"表示disperse卷,其中‘4’是数据,’2’是冗余校验,"volumetype: none"表示distribute卷;
allowVolumeExpansion:表示是否支持动态扩容,默认为true;
volumeBindingMode:表示是否立即bound pv,可选值WaitForFirstConsumer和Immediate;
创建sc:
# kubectl create -f storageclass-gfs-heketi-distributed.yaml
# kubectl get sc
NAME PROVISIONER RECLAIMPOLICY VOLUMEBINDINGMODE ALLOWVOLUMEEXPANSION
gluster-distributed kubernetes.io/glusterfs Retain Immediate true
创建pvc
vi glusterfs-pvc.yaml
apiVersion: v1
kind: PersistentVolumeClaim
metadata:
name: glusterfs-test
spec:
accessModes:
- ReadWriteMany
resources:
requests:
storage: 5Gi
storageClassName: gluster-distributed
创建pvc
# kubectl create -f glusterfs-pvc.yaml
# kubectl get pvc
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE
glusterfs-test Bound pvc-0a3cffea-36b0-4609-a983-88df995b99e8 5Gi RWX gluster-distributed 5s
这个时候 df -h 就可以看到有新的分区被挂载
# df -h | tail -n 1
/dev/mapper/vg_5b4ef948902b85845afd387ca71c6858-brick_0fd0afa344ca798d7a9fccab5bb60a9e 5.0G 33M 5.0G 1% /var/lib/heketi/mounts/vg_5b4ef948902b85845afd387ca71c6858/brick_0fd0afa344ca798d7a9fccab5bb60a9e
创建一个deployment,测试读写
vi glusterfs-deploy.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
name: glusterfs-deploy
labels:
app: nginx
spec:
replicas: 2
selector:
matchLabels:
app: nginx
template:
metadata:
labels:
app: nginx
spec:
containers:
- name: nginx
image: registry:5000/nginx:1.13
imagePullPolicy: IfNotPresent
ports:
- name: web
containerPort: 80
volumeMounts:
- name: pvc-glusterfs
mountPath: /data
volumes:
- name: pvc-glusterfs
persistentVolumeClaim:
claimName: glusterfs-test
创建deployment
# kubectl create -f glusterfs-deploy.yaml
deployment.apps/glusterfs-deploy created
# kubectl get pod | grep glus
glusterfs-deploy-cfd8c8578-45r9z 1/1 Running 0 72s
glusterfs-deploy-cfd8c8578-vjrzn 1/1 Running 0 104s
测试读写
# kubectl exec -it glusterfs-deploy-cfd8c8578-45r9z -- bash
# cd /data/
# for i in `seq 1 10`; do echo $i > $i ; done
# ls
1 10 2 3 4 5 6 7 8 9
# cat 1
1
至此,gfs已经可以作为k8s的storageclass来动态管理pv pvc了。
手机扫一扫
移动阅读更方便
你可能感兴趣的文章