本次教程目的是创建mysql 并挂载数据卷(StatefulSet + service + configmap + pvc)
基于kubernetes官方文档run-replicated-stateful-application、Kubernetes-部署高可用的MySQL
需要注意的是,本案例只是一个示范,不可实践于生产环境,仅用于理解StatefulSet以及tke操作实践
示例MySQL部署包括ConfigMap,PersistentVolumeClaim,两个Services和StatefulSet。
通过kubectl 创建 configmap
$ cat mysql-configmap.yaml
apiVersion: v1
kind: ConfigMap
metadata:
name: mysql
labels:
app: mysql
data:
master.cnf: |
[mysqld]
log-bin
log_bin_trust_function_creators=1
lower_case_table_names=1
slave.cnf: |
[mysqld]
super-read-only
log_bin_trust_function_creators=1
此ConfigMap提供了my.cnf覆盖,允许您独立控制MySQL主服务器和从服务器上的配置。在这种情况下,您希望主服务器能够将复制日志提供给从服务器,并且您希望从服务器拒绝任何不通过复制进行的写入。
ConfigMap本身并没有什么特别之处,它导致不同的部分应用于不同的Pod。每个Pod根据StatefulSet控制器提供的信息决定在初始化时查看哪个部分。
$ cat mysql-service.yaml
apiVersion: v1
kind: Service
metadata:
name: mysql
labels:
app: mysql
spec:
ports:
- name: mysql
port: 3306
clusterIP: None
selector:
app: mysql
---
apiVersion: v1
kind: Service
metadata:
name: mysql-read
labels:
app: mysql
spec:
ports:
- name: mysql
port: 3306
selector:
app: mysql
StatefulSet控制器为Pod创建了一个DNS条目,而Headless服务为DNS条目提供一个主机。因为Headless服务的名称为mysql,其他Pod通过<pod-name>.mysql访问此Pod。客户端访问被称为mysql-read,客户端服务通过访问mysql-read读取数据。通过连接myql执行写入数据的操作。
$ cat mysql-statefulset.yaml
apiVersion: apps/v1
kind: StatefulSet
metadata:
name: mysql
spec:
selector:
matchLabels:
app: mysql
serviceName: mysql
replicas: 3
template:
metadata:
labels:
app: mysql
spec:
initContainers:
- name: init-mysql
image: mysql:5.7
command:
- bash
- "-c"
- |
set -ex
[[ `hostname` =~ -([0-9]+)$ ]] || exit 1
ordinal=${BASH_REMATCH[1]}
echo [mysqld] > /mnt/conf.d/server-id.cnf
echo server-id=$((100 + $ordinal)) >> /mnt/conf.d/server-id.cnf
if [[ $ordinal -eq 0 ]]; then
cp /mnt/config-map/master.cnf /mnt/conf.d/
else
cp /mnt/config-map/slave.cnf /mnt/conf.d/
fi
volumeMounts:
- name: conf
mountPath: /mnt/conf.d
- name: config-map
mountPath: /mnt/config-map
- name: clone-mysql
image: ist0ne/xtrabackup
command:
- bash
- "-c"
- |
set -ex
[[ -d /var/lib/mysql/mysql ]] && exit 0
[[ `hostname` =~ -([0-9]+)$ ]] || exit 1
ordinal=${BASH_REMATCH[1]}
[[ $ordinal -eq 0 ]] && exit 0
ncat --recv-only mysql-$(($ordinal-1)).mysql 3307 | xbstream -x -C /var/lib/mysql
xtrabackup --prepare --target-dir=/var/lib/mysql
volumeMounts:
- name: data
mountPath: /var/lib/mysql
subPath: mysql
- name: conf
mountPath: /etc/mysql/conf.d
containers:
- name: mysql
image: mysql:5.7
env:
- name: MYSQL_ALLOW_EMPTY_PASSWORD
value: "1"
ports:
- name: mysql
containerPort: 3306
volumeMounts:
- name: data
mountPath: /var/lib/mysql
subPath: mysql
- name: conf
mountPath: /etc/mysql/conf.d
resources:
requests:
cpu: 500m
memory: 1Gi
livenessProbe:
exec:
command: ["mysqladmin", "ping"]
initialDelaySeconds: 30
periodSeconds: 10
timeoutSeconds: 5
readinessProbe:
exec:
command: ["mysql", "-h", "127.0.0.1", "-e", "SELECT 1"]
initialDelaySeconds: 5
periodSeconds: 2
timeoutSeconds: 1
- name: xtrabackup
image: ist0ne/xtrabackup
ports:
- name: xtrabackup
containerPort: 3307
command:
- bash
- "-c"
- |
set -ex
cd /var/lib/mysql
if [[ -f xtrabackup_slave_info ]]; then
mv xtrabackup_slave_info change_master_to.sql.in
rm -f xtrabackup_binlog_info
elif [[ -f xtrabackup_binlog_info ]]; then
[[ `cat xtrabackup_binlog_info` =~ ^(.*?)[[:space:]]+(.*?)$ ]] || exit 1
rm xtrabackup_binlog_info
echo "CHANGE MASTER TO MASTER_LOG_FILE='${BASH_REMATCH[1]}',\
MASTER_LOG_POS=${BASH_REMATCH[2]}" > change_master_to.sql.in
fi
if [[ -f change_master_to.sql.in ]]; then
echo "Waiting for mysqld to be ready (accepting connections)"
until mysql -h 127.0.0.1 -e "SELECT 1"; do sleep 1; done
echo "Initializing replication from clone position"
mv change_master_to.sql.in change_master_to.sql.orig
mysql -h 127.0.0.1 <<EOF
$(<change_master_to.sql.orig),
MASTER_HOST='mysql-0.mysql',
MASTER_USER='root',
MASTER_PASSWORD='',
MASTER_CONNECT_RETRY=10;
START SLAVE;
EOF
fi
exec ncat --listen --keep-open --send-only --max-conns=1 3307 -c \
"xtrabackup --backup --slave-info --stream=xbstream --host=127.0.0.1 --user=root"
volumeMounts:
- name: data
mountPath: /var/lib/mysql
subPath: mysql
- name: conf
mountPath: /etc/mysql/conf.d
resources:
requests:
cpu: 100m
memory: 100Mi
volumes:
- name: conf
emptyDir: {}
- name: config-map
configMap:
name: mysql
volumeClaimTemplates:
- metadata:
name: data
spec:
accessModes: ["ReadWriteOnce"]
resources:
requests:
storage: 10Gi
通过kubectl create 创建所有
$ ll
total 16
-rw-r--r-- 1 root root 280 Apr 12 11:34 mysql-configmap.yaml
-rw-r--r-- 1 root root 329 Apr 12 11:34 mysql-service.yaml
-rw-r--r-- 1 root root 4380 Apr 12 11:35 mysql-statefulset.yaml
$ kubectl create -f .
configmap "mysql" created
service "mysql" created
service "mysql-read" created
statefulset.apps "mysql" created
可以通过运行来查看启动进度:
$ kubectl get pods -l app=mysql --watch
NAME READY STATUS RESTARTS AGE
mysql-0 2/2 Running 0 1m
mysql-1 0/2 PodInitializing 0 26s
mysql-1 1/2 Running 0 27s
mysql-1 2/2 Running 0 29s
mysql-2 0/2 Pending 0 0s
mysql-2 0/2 Pending 0 0s
mysql-2 0/2 Pending 0 15s
mysql-2 0/2 Init:0/2 0 15s
mysql-2 0/2 Init:1/2 0 26s
mysql-2 0/2 Init:1/2 0 29s
mysql-2 0/2 PodInitializing 0 36s
mysql-2 1/2 Running 0 39s
mysql-2 2/2 Running 0 41s
可以看到status有 PodInitializing
,Init:0/2
,这是sts中声明的initContainers作用,这里直接引用官方文档的解释understanding-stateful-pod-initialization
通俗来说:在这里initContainers的作用是根据序数索引生成特殊的MySQL配置文件。确保在启动mysql容器前先通过init-mysql初始化配置文件。
脚本从Pod名称的结尾处获取并确定它的顺序索引,顺序索引通过hostname命令获取。然后,它会按照顺序保存在conf.d目录下的server-id.cnf文件中。此行为将StatefulSet控制器提供的唯一和稳定的身份标识转为mysql服务Id的域。在init-mysql容器中,脚本使用来自于ConfigMap中master.cnf或slave.cnf。
在此例子的拓扑关系中,存在一个MySQL master节点和多个MySQL slave节点,脚本简单的指派顺序0给主节点。这能够保证MySQL主节点在创建从节点之前就已经准备就绪。
为什么pod的名字是 mysql-0 mysql-1 mysql-2?
StatefulSet控制器按顺序索引按顺序启动一个Pod。它会等待,直到每个Pod报告准备就绪,然后开始启动下一个。 此外,控制器为每个Pod分配一个唯一,稳定的pod名称
<statefulset-name>-<ordinal-index>
。在这种情况下,导致pod命名mysql-0
,mysql-1
和mysql-2
。
一般来说,当一个新的Pod加入进来作为从节点时,必须假设MySQL master已经有关于它的数据。也假设slave副本的日志必须重新开始的。这些假设对于StatefulSet的扩缩容是很关键。
第二个初始化容器是clone-mysql,它在空的PersistentVolume上执行克隆从节点Pod的行为。这意味着它将从已在运行的Pod中拷贝数据,因此,它的当前状态能够与从master开始的副本节点一致。
MySQL自身并没有提供能够做到上述能力的机制,因此,此例子使用开源的Percona XtraBackup工具来实现。在克隆的过程中,为了对MySQL主节点影响的最小化,脚本会要求每一个新的Pod从顺序索引值小的Pod中进行克隆。这样做的原因是,StatefulSet控制器需要一直保证Pod N需要在Pod N+1之前准备就绪。
在初始化容器完成后,容器将正常运行。MySQL Pod由运行实际mysqld
服务的myqsl
容器组成,xtrabacekup
容器只是作为备份的工具。xtrabackup
负责监控克隆数据文件,并确定是否在从节点初始化MySQL副本。如果是的话,它将等待mysqld
就绪,然后执行 CHANGE MASTER TO
并START SLAVE
从XtraBackup
克隆文件中提取复制参数命令。
一旦一个从节点开始复制,它将记住MySQL master,并自动进行重新连接,因为从节点寻找主节点作为稳定DNS名称(mysql-0.mysql),它们自动的发现主节点。最后,在启动副本后,xtrabackup容器也监听来自于其它Pod对数据克隆的请求。
通过运行一个容器(mysql:5.7镜像),使用MySQL 客户端发送测试请求给MySQL master节点(主机名为mysql-0.mysql)
kubectl run mysql-client --image=mysql:5.7 -i --rm --restart=Never --\
mysql -h mysql-0.mysql <<EOF
CREATE DATABASE test;
CREATE TABLE test.messages (message VARCHAR(250));
INSERT INTO test.messages VALUES ('hello');
EOF
$ kubectl run mysql-client --image=mysql:5.7 -i --rm --restart=Never --\
> mysql -h mysql-0.mysql <<EOF
> CREATE DATABASE test;
> CREATE TABLE test.messages (message VARCHAR(250));
> INSERT INTO test.messages VALUES ('hello');
> EOF
If you don't see a command prompt, try pressing enter.
在master节点上创建demo数据库,并创建一个只有message字段的demo.messages的表,并为message字段插入hello值。
使用主机mysql-read
名将测试查询发送到报告为Ready的任何服务器:
kubectl run mysql-client --image=mysql:5.7 -i -t --rm --restart=Never --\
mysql -h mysql-read -e "SELECT * FROM test.messages"
$ kubectl run mysql-client --image=mysql:5.7 -i -t --rm --restart=Never --\
> mysql -h mysql-read -e "SELECT * FROM test.messages"
+---------+
| message |
+---------+
| hello |
+---------+
要演示mysql-read
服务在服务器之间分配连接,您可以SELECT @@server_id
在循环中运行:
kubectl run mysql-client-loop --image=mysql:5.7 -i -t --rm --restart=Never --\
bash -ic "while sleep 1; do mysql -h mysql-read -e 'SELECT @@server_id,NOW()'; done"
# kubectl run mysql-client-loop --image=mysql:5.7 -i -t --rm --restart=Never --\
> bash -ic "while sleep 1; do mysql -h mysql-read -e 'SELECT @@server_id,NOW()'; done"
If you don't see a command prompt, try pressing enter.
+-------------+---------------------+
| @@server_id | NOW() |
+-------------+---------------------+
| 101 | 2019-04-12 07:27:37 |
+-------------+---------------------+
+-------------+---------------------+
| @@server_id | NOW() |
+-------------+---------------------+
| 102 | 2019-04-12 07:27:38 |
+-------------+---------------------+
+-------------+---------------------+
| @@server_id | NOW() |
+-------------+---------------------+
| 100 | 2019-04-12 07:27:39 |
+-------------+---------------------+
^C
pod default/mysql-client-loop terminated (Error)
把mysql容器中的mysql重命名达到mysql -h 127.0.0.1 -e 'SELECT 1'
无法执行,达到容器异常效果
kubectl exec mysql-2 -c mysql -- mv /usr/bin/mysql /usr/bin/mysql.off
几秒钟后,Pod应该将其中一个容器报告为未就绪,您可以通过运行来检查:
$ kubectl get pod
NAME READY STATUS RESTARTS AGE
mysql-0 2/2 Running 0 40m
mysql-1 2/2 Running 0 39m
mysql-2 1/2 Running 0 39m
此时通过SELECT @@server_id
查看会发现102
消失不报告了,回想一下init-mysql
脚本定义server-id
为100 + $ordinal
,所以server ID 102
对应Pod mysql-2
+-------------+---------------------+
| @@server_id | NOW() |
+-------------+---------------------+
| 102 | 2019-04-12 07:30:43 |
+-------------+---------------------+
+-------------+---------------------+
| @@server_id | NOW() |
+-------------+---------------------+
| 102 | 2019-04-12 07:30:44 |
+-------------+---------------------+
+-------------+---------------------+
| @@server_id | NOW() |
+-------------+---------------------+
| 100 | 2019-04-12 07:30:45 |
+-------------+---------------------+
+-------------+---------------------+
| @@server_id | NOW() |
+-------------+---------------------+
| 101 | 2019-04-12 07:30:46 |
+-------------+---------------------+
+-------------+---------------------+
| @@server_id | NOW() |
+-------------+---------------------+
| 101 | 2019-04-12 07:30:47 |
+-------------+---------------------+
+-------------+---------------------+
| @@server_id | NOW() |
+-------------+---------------------+
| 101 | 2019-04-12 07:30:48 |
+-------------+---------------------+
+-------------+---------------------+
| @@server_id | NOW() |
+-------------+---------------------+
| 100 | 2019-04-12 07:30:49 |
+-------------+---------------------+
修复pod,几秒钟后应该重新出现在循环中
kubectl exec mysql-2 -c mysql -- mv /usr/bin/mysql.off /usr/bin/mysql
# kubectl run mysql-client-loop --image=mysql:5.7 -i -t --rm --restart=Never -- bash -ic "while sleep 1; do mysql -h mysql-read -e 'SELECT @@server_id,NOW()'; done"
If you don't see a command prompt, try pressing enter.
+-------------+---------------------+
| @@server_id | NOW() |
+-------------+---------------------+
| 101 | 2019-04-12 07:36:53 |
+-------------+---------------------+
+-------------+---------------------+
| @@server_id | NOW() |
+-------------+---------------------+
| 101 | 2019-04-12 07:36:54 |
+-------------+---------------------+
+-------------+---------------------+
| @@server_id | NOW() |
+-------------+---------------------+
| 101 | 2019-04-12 07:36:55 |
+-------------+---------------------+
+-------------+---------------------+
| @@server_id | NOW() |
+-------------+---------------------+
| 100 | 2019-04-12 07:36:56 |
+-------------+---------------------+
+-------------+---------------------+
| @@server_id | NOW() |
+-------------+---------------------+
| 101 | 2019-04-12 07:36:57 |
+-------------+---------------------+
+-------------+---------------------+
| @@server_id | NOW() |
+-------------+---------------------+
| 100 | 2019-04-12 07:36:58 |
+-------------+---------------------+
+-------------+---------------------+
| @@server_id | NOW() |
+-------------+---------------------+
| 102 | 2019-04-12 07:36:59 |
+-------------+---------------------+
如果Pods被删除,StatefulSet也会重新创建Pods,类似于ReplicaSet对无状态Pod的处理。
这里不做演示,同样的删除了pod,sts会重新拉起,并创建一个具有相同名称的pod并链接到pvc,可以通过SELECT @@server_id,NOW()
去观察102会消失一段时间又复原。
模拟节点故障
首先确定mysql-2运行在哪个节点上:
$ kubectl get pod -o wide
NAME READY STATUS RESTARTS AGE IP NODE
mysql-0 2/2 Running 0 53m 172.16.1.8 10.0.128.14
mysql-1 2/2 Running 0 53m 172.16.2.4 10.0.128.13
mysql-2 2/2 Running 0 52m 172.16.0.5 10.0.128.9
然后驱逐10.0.128.9
kubectl drain 10.0.128.9 --force --delete-local-data --ignore-daemonsets
$ kubectl drain 10.0.128.9 --force --delete-local-data --ignore-daemonsets
node "10.0.128.9" cordoned
WARNING: Deleting pods with local storage: mysql-2; Ignoring DaemonSet-managed pods: ip-masq-agent-gjtvn
pod "kube-dns-898dbbfc6-fh94g" evicted
pod "mysql-2" evicted
pod "kube-dns-898dbbfc6-jd42p" evicted
node "10.0.128.9" drained
驱逐完成后查看pod状态,可以看到mysql-2又被拉起来并在13节点上启动
$ kubectl get pod -o wide
NAME READY STATUS RESTARTS AGE IP NODE
mysql-0 2/2 Running 0 56m 172.16.1.8 10.0.128.14
mysql-1 2/2 Running 0 56m 172.16.2.4 10.0.128.13
mysql-2 0/2 Init:0/2 0 30s <none> 10.0.128.13
$ kubectl get pod -o wide
NAME READY STATUS RESTARTS AGE IP NODE
mysql-0 2/2 Running 0 57m 172.16.1.8 10.0.128.14
mysql-1 2/2 Running 0 56m 172.16.2.4 10.0.128.13
mysql-2 2/2 Running 0 1m 172.16.2.5 10.0.128.13
复原节点
$ kubectl uncordon 10.0.128.9
node "10.0.128.9" uncordoned
对于mysql副本,通过添加从节点进行扩容。
$ kubectl scale statefulset mysql --replicas=5
statefulset.apps "mysql" scaled
一旦它们启动,您应该看到服务器ID 103并104开始出现在SELECT @@server_id循环输出中。
$ kubectl get pod
NAME READY STATUS RESTARTS AGE
mysql-0 2/2 Running 0 1h
mysql-1 2/2 Running 0 59m
mysql-2 2/2 Running 0 4m
mysql-3 2/2 Running 0 1m
mysql-4 0/2 Init:1/2 0 31s
还可以验证这些新实例是否包含您在存在之前添加的数据,这里链接mysql-3
$ kubectl run mysql-client --image=mysql:5.7 -i -t --rm --restart=Never --\
> mysql -h mysql-3.mysql -e "SELECT * FROM test.messages"
+---------+
| message |
+---------+
| hello |
+---------+
kubectl scale statefulset mysql --replicas=3
$ kubectl scale statefulset mysql --replicas=3
statefulset.apps "mysql" scaled
但是请注意,虽然扩容后会自动创建新的pvc,但是缩容后不会自动删除这些pvc,你可以选择保留这些pvc,或手动删除,保留将产生费用请自行评估(用也会产生费用,具体的看storageclasses如何配置,TKE集群默认会有一个名为cbs的storageclasses,按量计费,云硬盘类型为普通云硬盘)
运行这个查看pvc:
kubectl get pvc -l app=mysql
$kubectl get pvc
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE
data-mysql-0 Bound pvc-518a14d5-5cef-11e9-b7b3-860056e03b4e 50Gi RWO cbs 1h
data-mysql-1 Bound pvc-6e00af5f-5cef-11e9-b7b3-860056e03b4e 50Gi RWO cbs 1h
data-mysql-2 Bound pvc-7f3d68cd-5cef-11e9-b7b3-860056e03b4e 50Gi RWO cbs 1h
data-mysql-3 Bound pvc-9b71a84c-5cf7-11e9-b7b3-860056e03b4e 50Gi RWO cbs 7m
data-mysql-4 Bound pvc-b4a1d96f-5cf7-11e9-b7b3-860056e03b4e 50Gi RWO cbs 6m
直接delete 即可,sts会一个一个回收,时间较长。
$ kubectl delete -f .
configmap "mysql" deleted
service "mysql" deleted
service "mysql-read" deleted
pvc需手动删除
到此结束。
本文系外文翻译,前往查看
如有侵权,请联系 cloudcommunity@tencent.com 删除。
本文系外文翻译,前往查看
如有侵权,请联系 cloudcommunity@tencent.com 删除。