13 副本机制和其他控制器

发表于 | 分类于


保持 pod 的健康

  1. 在实际的应用里，希望部署的pod能自动保持运行和健康，无需手动干预
  2. 不要直接创建pod，而是创建RC或者Deployment这样的资源由他们来管理
  3. 如何托管我们的pod，k8s如何自动重启pod，node节点失败pod如何被调度。
  4. 即使进程崩溃，应用程序会停止工作，比如java进程的内存泄露

存活探针

1
2
3
4
5
6
7
8
# k8s三种探测容器机制 

1. httpGet
发送HTTP请求，返回2XX-3XX范围状态码为成功。
2. exec
执行Shell命令返回状态码是0为成功。
3. tcpSocket
发起TCP Socket建立成功。

基于HTTP的存活探针

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
# https://github.com/luksa/kubernetes-in-action/tree/master/Chapter04/kubia-unhealthy

[root@k8s-master1 demo]# vim kubia-libeness-probe.yaml 

apiVersion: v1
kind: Pod
metadata:
  name: kubia-liveness
spec:
  containers:
  - image: 172.31.228.68/project/kubia-httpget   # 出现访问问题的镜像
    name: kubia
    livenessProbe: # 存活探针
      httpGet:     # httpget类型
        path: /    # 请求路径
        port: 8080 # 请求端口

# 该镜像出现2次访问后会返回 http 500
1
2
3
4
5
6
7
8
9
10
11
12
# 大约1分半后容器会重启
# RESTARTS 显示重启次数
# 现在的模式是 无限循环的重启
[root@k8s-master1 demo]# kubectl create -f kubia-libeness-probe.yaml 
pod/kubia-liveness created

[root@k8s-master1 demo]# kubectl get pods
NAME             READY   STATUS    RESTARTS   AGE
kubia-liveness   1/1     Running   0          61s
[root@k8s-master1 demo]# kubectl get pods
NAME             READY   STATUS    RESTARTS   AGE
kubia-liveness   1/1     Running   1          2m1s

获取崩溃容器的应用日志

1
2
3
4
5
6
7
8
9
10
11
12
# 当想知道前一个容器日志,而不是当前的容器时 加上 --previous
[root@k8s-master1 demo]# kubectl logs kubia-liveness --previous

Kubia server starting...
Received request from ::ffff:10.244.2.1
Received request from ::ffff:10.244.2.1
Received request from ::ffff:10.244.2.1
Received request from ::ffff:10.244.2.1
Received request from ::ffff:10.244.2.1
Received request from ::ffff:10.244.2.1
Received request from ::ffff:10.244.2.1
Received request from ::ffff:10.244.2.1

查看事件

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
[root@k8s-master1 demo]# kubectl describe pod kubia-liveness

# 先前容器发生错误 返回码 137
# 137 = 128+x 其中x是终止进程的信号编号,在这个例子里x=9，是sigkill的信号编号,意味着被强行终止
# Restart Count 重启次数
...
    Last State:     Terminated
      Reason:       Error
      Exit Code:    137
      Started:      Wed, 11 Mar 2020 11:39:53 +0800
      Finished:     Wed, 11 Mar 2020 11:41:43 +0800
    Ready:          True
    Restart Count:  3
    Liveness:       http-get http://:8080/ delay=0s timeout=1s period=10s #success=1 #failure=3
...


Events:
  Type     Reason     Age                    From                  Message
  ----     ------     ----                   ----                  -------
  Normal   Scheduled  <unknown>              default-scheduler     Successfully assigned default/kubia-liveness to k8s-master1
  Normal   Created    4m36s (x3 over 8m11s)  kubelet, k8s-master1  Created container kubia
  Normal   Started    4m36s (x3 over 8m11s)  kubelet, k8s-master1  Started container kubia
  Warning  Unhealthy  3m16s (x9 over 7m16s)  kubelet, k8s-master1  Liveness probe failed: HTTP probe failed with statuscode: 500
  Normal   Killing    3m16s (x3 over 6m56s)  kubelet, k8s-master1  Container kubia failed liveness probe, will be restarted
  Normal   Pulling    2m46s (x4 over 8m11s)  kubelet, k8s-master1  Pulling image "172.31.228.68/project/kubia-httpget"
  Normal   Pulled     2m46s (x4 over 8m11s)  kubelet, k8s-master1  Successfully pulled image "172.31.228.68/project/kubia-httpget"

# 事件中告诉我们为什么会被重启
# 当容器被强行终止时,会创建一个全新的容器,而不是重启原来的容器

配置存活探针的附加属性

1
2
3
4
5
6
7
8
9
10
11
12
# http-get http://:8080/ delay=0s timeout=1s period=10s #success=1 #failure=3
# delay 延迟  
# timeout 超时
# period 周期

# delay=0s 容器启动后立即探测
# timeout=1s 容器需要再1秒钟内响应，否则检测失败
# period=10s 每10秒探测一次
# failure=3 连续3次探测后重启容器

# 务必设置一个初始延迟来说明应用的启动时间
# 退出代码 137表示进程被外部信号终止,128+9  如果是143对应128+15
1
2
# initialDelaySeconds: 10 # 容器启动5秒后 开始健康检查
# periodSeconds: 5        # 每隔5秒执行一次
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
[root@k8s-master1 demo]# vim kubia-libeness-probe.yaml 

apiVersion: v1
kind: Pod
metadata:
  name: kubia-liveness
spec:
  containers:
  - image: 172.31.228.68/project/kubia-httpget
    name: kubia
    livenessProbe:
      httpGet:
        path: /
        port: 8080
      initialDelaySeconds: 10
      periodSeconds: 5

[root@k8s-master1 demo]# kubectl describe pod kubia-liveness
...
 Liveness:       http-get http://:8080/ delay=10s timeout=1s period=5s #success=1 #failure=3
...

创建 ReplicationController

1
2
3
4
# ReplicationController 三部分
# 1. 标签选择器
# 2. 副本个数
# 3. pod 模板
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
[root@k8s-master1 demo]# vim kubia-rc.yaml 

apiVersion: v1
kind: ReplicationController        # 资源对象
metadata:
  name: kubia                      # rc名称
spec:
  replicas: 3                      # 副本个数
  selector:
    app: kubia                     # pod 选择器
  template:                        # pod 模板
    metadata:
      labels:
        app: kubia                 # 模板中的pod标签 需要与pod标签选择器相同
    spec:
      containers:
      - image: 172.31.228.68/project/kubia
        name: kubia
        ports:
        - containerPort: 8080
        protocol: TCP

[root@k8s-master1 demo]# kubectl create -f kubia-rc.yaml 

[root@k8s-master1 demo]# kubectl get pods
NAME          READY   STATUS    RESTARTS   AGE
kubia-dhj7g   1/1     Running   0          33s
kubia-k659d   1/1     Running   0          33s
kubia-l4fpp   1/1     Running   0          33s

# 删除pod 再查看
[root@k8s-master1 demo]# kubectl delete pod kubia-dhj7g

[root@k8s-master1 ~]# kubectl get pods
NAME          READY   STATUS        RESTARTS   AGE
kubia-dhj7g   1/1     Terminating   0          115s        # 终止
kubia-k659d   1/1     Running       0          115s
kubia-l4fpp   1/1     Running       0          115s
kubia-mqk8k   1/1     Running       0          28s         # 重新创建

# 查看 rc
[root@k8s-master1 demo]# kubectl get rc
NAME    DESIRED   CURRENT   READY   AGE
kubia   3         3         3       2m35s

# 查看rc详细信息
[root@k8s-master1 demo]# kubectl describe rc kubia
Name:         kubia
Namespace:    default
Selector:     app=kubia
Labels:       app=kubia
Annotations:  <none>
Replicas:     3 current / 3 desired                            # pod 实际数量和目标数量
Pods Status:  3 Running / 0 Waiting / 0 Succeeded / 0 Failed   # 每种状态下的pod数量
Pod Template:
  Labels:  app=kubia
  Containers:
   kubia:
    Image:        172.31.228.68/project/kubia
    Port:         8080/TCP
    Host Port:    0/TCP
    Environment:  <none>
    Mounts:       <none>
  Volumes:        <none>
Events:
  Type    Reason            Age    From                    Message
  ----    ------            ----   ----                    -------
  Normal  SuccessfulCreate  7m51s  replication-controller  Created pod: kubia-dhj7g
  Normal  SuccessfulCreate  7m51s  replication-controller  Created pod: kubia-l4fpp
  Normal  SuccessfulCreate  7m51s  replication-controller  Created pod: kubia-k659d
  Normal  SuccessfulCreate  6m24s  replication-controller  Created pod: kubia-mqk8k

断开 node2 节点测试

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
[root@k8s-master1 demo]# kubectl get pods -o wide
NAME          READY   STATUS    RESTARTS   AGE   IP            NODE          NOMINATED NODE   READINESS GATES
kubia-k659d   1/1     Running   0          11m   10.244.1.16   k8s-node2     <none>           <none>
kubia-l4fpp   1/1     Running   0          11m   10.244.2.16   k8s-master1   <none>           <none>
kubia-mqk8k   1/1     Running   0          10m   10.244.0.14   k8s-node1     <none>           <none>

# 如果节点在几分钟之内无法访问,那么节点上的pod的状态会变成unknown，rc会重新创建pod
[root@k8s-master1 demo]# kubectl get nodes
NAME          STATUS     ROLES    AGE   VERSION
k8s-master1   Ready      <none>   46h   v1.16.0
k8s-node1     Ready      <none>   46h   v1.16.0
k8s-node2     NotReady   <none>   46h   v1.16.0

[root@k8s-master1 demo]# kubectl get pods -o wide
NAME          READY   STATUS        RESTARTS   AGE   IP            NODE          NOMINATED NODE   READINESS GATES
kubia-k659d   1/1     Terminating   0          18m   10.244.1.16   k8s-node2     <none>           <none>
kubia-l4fpp   1/1     Running       0          18m   10.244.2.16   k8s-master1   <none>           <none>
kubia-mqk8k   1/1     Running       0          16m   10.244.0.14   k8s-node1     <none>           <none>
kubia-sfhkc   1/1     Running       0          17s   10.244.2.17   k8s-master1   <none>           <none>

pod 迁移或迁出 rc 的作用域

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
1. 通过修改pod的标签 可以将它从rc的作用域中添加或者删除

# 给现有rc中的pod增加新标签
# rc并不关心 新增的标签 只关心pod是否有标签选择器中的标签 
[root@k8s-master1 demo]# kubectl get pods
NAME          READY   STATUS    RESTARTS   AGE
kubia-l4fpp   1/1     Running   0          29m
kubia-mqk8k   1/1     Running   0          27m
kubia-sfhkc   1/1     Running   0          11m

[root@k8s-master1 demo]# kubectl label pod kubia-l4fpp team=A
pod/kubia-l4fpp labeled

[root@k8s-master1 demo]# kubectl get pods --show-labels
NAME          READY   STATUS    RESTARTS   AGE   LABELS
kubia-l4fpp   1/1     Running   0          29m   app=kubia,team=A
kubia-mqk8k   1/1     Running   0          28m   app=kubia
kubia-sfhkc   1/1     Running   0          11m   app=kubia
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
# 更改已托管到rc中的pod标签

[root@k8s-master1 demo]# kubectl label pod kubia-l4fpp app=foo --overwrite
pod/kubia-l4fpp labeled

[root@k8s-master1 demo]# kubectl get pods --show-labels
NAME          READY   STATUS    RESTARTS   AGE   LABELS
kubia-l4fpp   1/1     Running   0          31m   app=foo,team=A  # 不再被rc管理，可以手动删除该pod
kubia-mqk8k   1/1     Running   0          30m   app=kubia
kubia-sfhkc   1/1     Running   0          13m   app=kubia
kubia-wsqgg   1/1     Running   0          4s    app=kubia       # rc新建的pod 

[root@k8s-master1 demo]# kubectl delete pod kubia-l4fpp

# 如果修改rc控制器的标签选择器，那么rc之前管理的pod都会被脱离该rc，并且会创建3个新的pod
# 不要修改rc控制器的标签选择器，可以修改pod模板

kubectl edit 修改模板

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
[root@k8s-master1 demo]# kubectl edit rc kubia

...
  template:
    metadata:
      creationTimestamp: null
      labels:
        app: kubia
        team: C        # 新增标签
...

# 1. 不会影响现有pod
# 2. 如果删除再重新创建 则会带有新标签
# 3. 如果我们修改了容器镜像，则相当于更新本次pod，后续还有更好的方法更新pod

[root@k8s-master1 demo]# kubectl get pods
NAME          READY   STATUS    RESTARTS   AGE
kubia-mqk8k   1/1     Running   0          39m
kubia-sfhkc   1/1     Running   0          22m
kubia-wsqgg   1/1     Running   0          9m2s

[root@k8s-master1 demo]# kubectl delete pod kubia-sfhkc kubia-wsqgg
pod "kubia-sfhkc" deleted
pod "kubia-wsqgg" deleted

[root@k8s-master1 ~]# kubectl get pods --show-labels
NAME          READY   STATUS    RESTARTS   AGE   LABELS
kubia-2ftc2   1/1     Running   0          40s   app=kubia,team=C
kubia-mqk8k   1/1     Running   0          40m   app=kubia
kubia-x6xm2   1/1     Running   0          40s   app=kubia,team=C

kubectl sacle 水平扩容

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
[root@k8s-master1 demo]# kubectl scale rc kubia --replicas=5
replicationcontroller/kubia scaled

[root@k8s-master1 demo]# kubectl get pods --show-labels
NAME          READY   STATUS    RESTARTS   AGE     LABELS
kubia-2ftc2   1/1     Running   0          4m18s   app=kubia,team=C
kubia-cf627   1/1     Running   0          32s     app=kubia,team=C
kubia-mqk8k   1/1     Running   0          44m     app=kubia
kubia-t2k4h   1/1     Running   0          32s     app=kubia,team=C
kubia-x6xm2   1/1     Running   0          4m18s   app=kubia,team=C

# edit 编辑定义修改 rc 
[root@k8s-master1 demo]# kubectl edit rc kubia
...
spec:
  replicas: 3  # 副本个数
...

[root@k8s-master1 demo]# kubectl get pods --show-labels
NAME          READY   STATUS    RESTARTS   AGE     LABELS
kubia-2ftc2   1/1     Running   0          7m30s   app=kubia,team=C
kubia-mqk8k   1/1     Running   0          47m     app=kubia
kubia-x6xm2   1/1     Running   0          7m30s   app=kubia,team=C

# 声明式的集群伸缩: 我想要运行x个实例

删除 ReplicationController

1
2
3
4
5
6
7
8
9
10
11
# 1. 删除rc的同时，它管理的pod也会被删除
# 2. 当删除rc时可以通过 --cascade=false 来保持pod的运行
# 3. 当你删除rc时,pod已经不受管理,可以新建rc管理他们,只要rc的标签管理器对应pod的标签

[root@k8s-master1 demo]# kubectl delete rc kubia --cascade=false

[root@k8s-master1 demo]# kubectl get pods --show-labels
NAME          READY   STATUS    RESTARTS   AGE   LABELS
kubia-2ftc2   1/1     Running   0          13m   app=kubia,team=C
kubia-mqk8k   1/1     Running   0          53m   app=kubia
kubia-x6xm2   1/1     Running   0          13m   app=kubia,team=C

使用 ReplicatSet 替换 ReplicationController

1
2
3
# 1. ReplicationController 是用于复制和在异常时重新调度节点的最早组件
# 2. ReplicatSet 是 新一代的 ReplicationController 并且已经完全替换掉
# 3. 通常不会直接创建ReplicatSet，而是创建更高层级的 Deployment时自动创建

ReplicatSet 和 ReplicationController 的区别

1
2
3
# 1. ReplicationController 只能匹配一组标签
# 2. ReplicatSet 可以匹配多组标签
# 3. ReplicatSet 可以匹配标签的键,比如名为env的标签,而不在乎值,(env=*)

定义 ReplicatSet

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
# 参考
[root@k8s-master1 demo]# kubectl explain rs
KIND:     ReplicaSet
VERSION:  apps/v1

[root@k8s-master1 demo]# kubectl explain rc
KIND:     ReplicationController
VERSION:  v1


# 创建 ReplicatSet 并且接管之前三个被移除rc的pod

[root@k8s-master1 demo]# vim kubia-rs.yaml

apiVersion: apps/v1
kind: ReplicaSet
metadata:
  name: kubia
spec:
  replicas: 3
  selector:
    matchLabels:       # 选择器里增加了 matchLabels选择器
      app: kubia       # 标签选择器
  template:
    metadata:
      labels:          # pod中的标签对应rs的标签选择器
        app: kubia
    spec:
      containers:
      - image: 172.31.228.68/project/kubia
        name: kubia
        ports:
        - containerPort: 8080

[root@k8s-master1 demo]# kubectl create -f kubia-rs.yaml 
replicaset.apps/kubia created

[root@k8s-master1 demo]# kubectl get rs
NAME    DESIRED   CURRENT   READY   AGE
kubia   3         3         3       26s

ReplicatSet 的标签选择器

1
2
3
4
5
6
7
8
9
10
11
12
# ReplicatSet 对于 rc的主要改进就是标签选择器

selector:
    matchExpressions:
      - key: app
        operator: In
        values:
         - kubia
# 会有四种 In NotIn Exists DoesNotExist 
# 如果是多个表达式 则所有条件必须都为true
# 如果同时指定 matchExpressions 和 matchLabels 则所有标签条件都必须为true
# 后续应该始终使用rs,当然也有其他人的部署里看到rc

删除 ReplicatSet

1
2
3
# 删除rs,则下面的pod同时删除
[root@k8s-master1 ~]# kubectl delete rs kubia
replicaset.apps "kubia" deleted

DaemonSet 在每个 node节点上运行1个pod

1
2
3
4
1. 如果 node节点下线 DaemonSet不会再其他节点上重新部署pod
2. 如果一个新的node节点上线,DaemonSet会立即在这个新节点上创建一个pod 
3. 除非指定pod在某个node上运行 nodeSelector 
4. DaemonSet 会绕过调度器，即使节点被设置为不可调度

创建 DaemonSet

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
[root@k8s-master2 ssd]# vim Dockerfile

FROM busybox
ENTRYPOINT while true; do echo 'SSD OK'; sleep 5; done

docker build -t ssd-monitor .
docker images
docker tag ssd-monitor 172.31.228.68/project/ssd-monitor
docker push 172.31.228.68/project/ssd-monitor

[root@k8s-master1 demo]# kubectl explain ds
KIND:     DaemonSet
VERSION:  apps/v1

[root@k8s-master1 demo]# vim kubia-ds.yaml 

apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: ssd-monitor
spec:
  selector:
    matchLabels:
      app: ssd-monitor
  template:
    metadata:
      labels:
        app: ssd-monitor
    spec:
      nodeSelector:            # 节点选择器 选择标签有disk: ssd的node
        disk: ssd
      containers:
      - image: 172.31.228.68/project/ssd-monitor
        name: main
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
[root@k8s-master1 demo]# kubectl create -f kubia-ds.yaml 

# 还没有给node 打上 disk=ssd标签 所有没有pod产生
[root@k8s-master1 demo]# kubectl get ds
NAME          DESIRED   CURRENT   READY   UP-TO-DATE   AVAILABLE   NODE SELECTOR   AGE
ssd-monitor   0         0         0       0            0           disk=ssd        33s

# 给两个node打上标签
[root@k8s-master1 demo]# kubectl get nodes
NAME          STATUS   ROLES    AGE   VERSION
k8s-master1   Ready    <none>   2d    v1.16.0
k8s-node1     Ready    <none>   2d    v1.16.0
k8s-node2     Ready    <none>   2d    v1.16.0

[root@k8s-master1 demo]# kubectl label node k8s-node1 k8s-node2 disk=ssd
node/k8s-node1 labeled
node/k8s-node2 labeled

[root@k8s-master1 demo]# kubectl get node -L disk
NAME          STATUS   ROLES    AGE   VERSION   DISK
k8s-master1   Ready    <none>   2d    v1.16.0   
k8s-node1     Ready    <none>   2d    v1.16.0   ssd
k8s-node2     Ready    <none>   2d    v1.16.0   ssd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
# 查看
[root@k8s-master1 demo]# kubectl get ds
NAME          DESIRED   CURRENT   READY   UP-TO-DATE   AVAILABLE   NODE SELECTOR   AGE
ssd-monitor   2         2         2       2            2           disk=ssd        3m29s

[root@k8s-master1 demo]# kubectl get pods -o wide
NAME                READY   STATUS    RESTARTS   AGE   IP            NODE        NOMINATED NODE   READINESS GATES
ssd-monitor-9twms   1/1     Running   0          4s    10.244.1.22   k8s-node2   <none>           <none>
ssd-monitor-tlrfq   1/1     Running   0          4s    10.244.0.16   k8s-node1   <none>           <none>

[root@k8s-master1 demo]# kubectl logs ssd-monitor-9twms
SSD OK
SSD OK
...

# 更换一个node标签 合理的下线了 ds

[root@k8s-master1 demo]# kubectl label node k8s-node2 disk=hdd --overwrite
node/k8s-node2 labeled

[root@k8s-master1 demo]# kubectl get pods -o wide
NAME                READY   STATUS        RESTARTS   AGE   IP            NODE        NOMINATED NODE   READINESS GATES
ssd-monitor-9twms   1/1     Terminating   0          74s   10.244.1.22   k8s-node2   <none>           <none>
ssd-monitor-tlrfq   1/1     Running       0          74s   10.244.0.16   k8s-node1   <none>           <none>

[root@k8s-master1 demo]# kubectl get ds
NAME          DESIRED   CURRENT   READY   UP-TO-DATE   AVAILABLE   NODE SELECTOR   AGE
ssd-monitor   1         1         1       1            1           disk=ssd        77s
1
2
3
# 删除 ds
[root@k8s-master1 demo]# kubectl delete ds ssd-monitor
daemonset.apps "ssd-monitor" deleted

Job 运行单个任务的 pod

1
2
3
1. jod 一旦运行完成不会重启容器
2. Job 分为 普通任务(Job) 和 定时任务 (CronJob) 
3. 应用场景: 离线数据处理，视频解码等业务

创建 Job

1
2
3
4
5
6
7
8
[root@k8s-master2 job]# vim Dockerfile 

FROM busybox
ENTRYPOINT echo "$(date) Batch job starting"; sleep 120; echo "$(date) Finished succesfully"

[root@k8s-master2 job]# docker build -t batch-job .
[root@k8s-master2 job]# docker tag batch-job 172.31.228.68/project/batch-job
[root@k8s-master2 job]# docker push 172.31.228.68/project/batch-job
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
[root@k8s-master1 demo]# kubectl explain job
KIND:     Job
VERSION:  batch/v1

[root@k8s-master1 demo]# vim kubia-job.yaml 

apiVersion: batch/v1
kind: Job
metadata:
  name: batch-job
spec:
  template:
    metadata:
      labels:
        app: batch-job
    spec:
      restartPolicy: OnFailure
      containers:
      - image: 172.31.228.68/project/batch-job
        name: main

[root@k8s-master1 demo]# kubectl create -f kubia-job.yaml
1
2
3
• Always：   当容器终止退出后，总是重启容器，默认策略。
• OnFailure：当容器异常退出（退出状态码非0）时，才重启容器。
• Never:：   当容器终止推出，从不重启容器。

查看 Job 运行

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
[root@k8s-master1 demo]# kubectl get jobs
NAME        COMPLETIONS   DURATION   AGE
batch-job   0/1           50s        50s


[root@k8s-master1 demo]# kubectl get pods
NAME              READY   STATUS    RESTARTS   AGE
batch-job-5tr5g   1/1     Running   0          70s

[root@k8s-master1 demo]# kubectl describe jobs batch-job

[root@k8s-master1 demo]# kubectl get pods
NAME              READY   STATUS      RESTARTS   AGE
batch-job-5tr5g   0/1     Completed   0          2m31s

# 读取 job 日志
[root@k8s-master1 demo]# kubectl logs batch-job-5tr5g 
Wed Mar 11 10:00:51 UTC 2020 Batch job starting
Wed Mar 11 10:02:51 UTC 2020 Finished succesfully

# 删除 job
[root@k8s-master1 demo]# kubectl delete job batch-job
job.batch "batch-job" deleted

计划任务 CronJob

创建 CronJob

1
2
3
4
5
6
7
8
[root@k8s-master1 demo]# kubectl explain cronjob
KIND:     CronJob
VERSION:  batch/v1beta1

定时任务，像Linux的Crontab一样
应用场景: 通知,备份
crontab的格式如下：
分 时 日 月 周 要运行的命令: 第1列分钟0～59 第2列小时0～23） 第3列日1～31 第4列月1～12 第5列星期0～7（0和7表示星期天） 第6列要运行的命令
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
[root@k8s-master1 demo]# vim cronjob.yaml 

apiVersion: batch/v1beta1
kind: CronJob
metadata:
  name: hello
spec:
  schedule: "*/1 * * * *"
  jobTemplate:
    spec:
      template:
        spec:
          containers:
          - name: hello
            image: busybox
            args:
            - /bin/sh
            - -c
            - date; echo Hello from the Kubernetes cluster
          restartPolicy: OnFailure


[root@k8s-master1 demo]# kubectl get cronjob
NAME    SCHEDULE      SUSPEND   ACTIVE   LAST SCHEDULE   AGE
hello   */1 * * * *   False     0        <none>          12s

[root@k8s-master1 demo]# kubectl get cronjob
NAME    SCHEDULE      SUSPEND   ACTIVE   LAST SCHEDULE   AGE
hello   */1 * * * *   False     0        31s             58s

[root@k8s-master1 demo]# kubectl get pods
NAME                     READY   STATUS      RESTARTS   AGE
hello-1583921700-s4rwp   0/1     Completed   0          28s

[root@k8s-master1 demo]# kubectl logs hello-1583921700-s4rwp
Wed Mar 11 10:15:16 UTC 2020
Hello from the Kubernetes cluster

# 到达定时后 会再次执行并产生pod

[root@k8s-master1 demo]# kubectl get pods
NAME                     READY   STATUS              RESTARTS   AGE
hello-1583921700-s4rwp   0/1     Completed           0          61s
hello-1583921760-sxz8q   0/1     ContainerCreating   0          1s

# 删除 cronjob
[root@k8s-master1 demo]# kubectl delete cronjob hello