amazon-web-services - `the node was low on resource imagefs` -- 定期导致 pod 驱逐

Question

我正在将 Jenkins-X 用于一个相对较大的项目，该项目由大约 30 个模块组成，其中 15 个是服务(因此，包含 Dockerfile 和一个用于部署的相应 Helm 图表)。

在其中一些相对较大的构建中，我间歇性地(~每隔一个构建)看到构建 pod 被驱逐，使用 kubectl describe pod <podname>我可以进行调查，我注意到 pod 因以下原因被驱逐:
the node was low on resource imagefs
完整数据:

Name:         maven-96wmn
Namespace:    jx
Node:         ip-192-168-66-176.eu-west-1.compute.internal/
Start Time:   Tue, 06 Nov 2018 10:22:54 +0000
Labels:       jenkins=slave
              jenkins/jenkins-maven=true
Annotations:  <none>
Status:       Failed
Reason:       Evicted
Message:      The node was low on resource: imagefs.
IP:           
Containers:
  maven:
    Image:      jenkinsxio/builder-maven:0.0.516
    Port:       <none>
    Host Port:  <none>
    Command:
      /bin/sh
      -c
    Args:
      cat
    Limits:
      cpu:     1
      memory:  1Gi
    Requests:
      cpu:     400m
      memory:  512Mi
    Environment:
      JENKINS_SECRET:       131c407141521c0842f62a69004df926be6cb531f9318edf0885aeb96b0662b4
      JENKINS_TUNNEL:       jenkins-agent:50000
      DOCKER_CONFIG:        /home/jenkins/.docker/
      GIT_AUTHOR_EMAIL:     jenkins-x@googlegroups.com
      GIT_COMMITTER_EMAIL:  jenkins-x@googlegroups.com
      GIT_COMMITTER_NAME:   jenkins-x-bot
      _JAVA_OPTIONS:        -XX:+UnlockExperimentalVMOptions -XX:+UseCGroupMemoryLimitForHeap -Dsun.zip.disableMemoryMapping=true -XX:+UseParallelGC -XX:MinHeapFreeRatio=5 -XX:MaxHeapFreeRatio=10 -XX:GCTimeRatio=4 -XX:AdaptiveSizePolicyWeight=90 -Xms10m -Xmx192m
      GIT_AUTHOR_NAME:      jenkins-x-bot
      JENKINS_NAME:         maven-96wmn
      XDG_CONFIG_HOME:      /home/jenkins
      JENKINS_URL:          http://jenkins:8080
      HOME:                 /home/jenkins
    Mounts:
      /home/jenkins from workspace-volume (rw)
      /home/jenkins/.docker from volume-2 (rw)
      /home/jenkins/.gnupg from volume-3 (rw)
      /root/.m2 from volume-1 (rw)
      /var/run/docker.sock from volume-0 (rw)
      /var/run/secrets/kubernetes.io/serviceaccount from jenkins-token-smvvp (ro)
  jnlp:
    Image:      jenkinsci/jnlp-slave:3.14-1
    Port:       <none>
    Host Port:  <none>
    Args:
      131c407141521c0842f62a69004df926be6cb531f9318edf0885aeb96b0662b4
      maven-96wmn
    Requests:
      cpu:     100m
      memory:  128Mi
    Environment:
      JENKINS_SECRET:       131c407141521c0842f62a69004df926be6cb531f9318edf0885aeb96b0662b4
      JENKINS_TUNNEL:       jenkins-agent:50000
      DOCKER_CONFIG:        /home/jenkins/.docker/
      GIT_AUTHOR_EMAIL:     jenkins-x@googlegroups.com
      GIT_COMMITTER_EMAIL:  jenkins-x@googlegroups.com
      GIT_COMMITTER_NAME:   jenkins-x-bot
      _JAVA_OPTIONS:        -XX:+UnlockExperimentalVMOptions -XX:+UseCGroupMemoryLimitForHeap -Dsun.zip.disableMemoryMapping=true -XX:+UseParallelGC -XX:MinHeapFreeRatio=5 -XX:MaxHeapFreeRatio=10 -XX:GCTimeRatio=4 -XX:AdaptiveSizePolicyWeight=90 -Xms10m -Xmx192m
      GIT_AUTHOR_NAME:      jenkins-x-bot
      JENKINS_NAME:         maven-96wmn
      XDG_CONFIG_HOME:      /home/jenkins
      JENKINS_URL:          http://jenkins:8080
      HOME:                 /home/jenkins
    Mounts:
      /home/jenkins from workspace-volume (rw)
      /home/jenkins/.docker from volume-2 (rw)
      /home/jenkins/.gnupg from volume-3 (rw)
      /root/.m2 from volume-1 (rw)
      /var/run/docker.sock from volume-0 (rw)
      /var/run/secrets/kubernetes.io/serviceaccount from jenkins-token-smvvp (ro)
Volumes:
  volume-0:
    Type:          HostPath (bare host directory volume)
    Path:          /var/run/docker.sock
    HostPathType:  
  volume-2:
    Type:        Secret (a volume populated by a Secret)
    SecretName:  jenkins-docker-cfg
    Optional:    false
  volume-1:
    Type:        Secret (a volume populated by a Secret)
    SecretName:  jenkins-maven-settings
    Optional:    false
  workspace-volume:
    Type:    EmptyDir (a temporary directory that shares a pod's lifetime)
    Medium:  
  volume-3:
    Type:        Secret (a volume populated by a Secret)
    SecretName:  jenkins-release-gpg
    Optional:    false
  jenkins-token-smvvp:
    Type:        Secret (a volume populated by a Secret)
    SecretName:  jenkins-token-smvvp
    Optional:    false
QoS Class:       Burstable
Node-Selectors:  <none>
Tolerations:     node.kubernetes.io/not-ready:NoExecute for 300s
                 node.kubernetes.io/unreachable:NoExecute for 300s
Events:
  Type     Reason                 Age   From                                                   Message
  ----     ------                 ----  ----                                                   -------
  Normal   Created                7m    kubelet, ip-192-168-66-176.eu-west-1.compute.internal  Created container
  Normal   SuccessfulMountVolume  7m    kubelet, ip-192-168-66-176.eu-west-1.compute.internal  MountVolume.SetUp succeeded for volume "workspace-volume"
  Normal   SuccessfulMountVolume  7m    kubelet, ip-192-168-66-176.eu-west-1.compute.internal  MountVolume.SetUp succeeded for volume "volume-0"
  Normal   SuccessfulMountVolume  7m    kubelet, ip-192-168-66-176.eu-west-1.compute.internal  MountVolume.SetUp succeeded for volume "volume-1"
  Normal   SuccessfulMountVolume  7m    kubelet, ip-192-168-66-176.eu-west-1.compute.internal  MountVolume.SetUp succeeded for volume "volume-2"
  Normal   SuccessfulMountVolume  7m    kubelet, ip-192-168-66-176.eu-west-1.compute.internal  MountVolume.SetUp succeeded for volume "volume-3"
  Normal   SuccessfulMountVolume  7m    kubelet, ip-192-168-66-176.eu-west-1.compute.internal  MountVolume.SetUp succeeded for volume "jenkins-token-smvvp"
  Normal   Pulled                 7m    kubelet, ip-192-168-66-176.eu-west-1.compute.internal  Container image "jenkinsxio/builder-maven:0.0.516" already present on machine
  Normal   Scheduled              7m    default-scheduler                                      Successfully assigned maven-96wmn to ip-192-168-66-176.eu-west-1.compute.internal
  Normal   Started                7m    kubelet, ip-192-168-66-176.eu-west-1.compute.internal  Started container
  Normal   Pulled                 7m    kubelet, ip-192-168-66-176.eu-west-1.compute.internal  Container image "jenkinsci/jnlp-slave:3.14-1" already present on machine
  Normal   Created                7m    kubelet, ip-192-168-66-176.eu-west-1.compute.internal  Created container
  Normal   Started                7m    kubelet, ip-192-168-66-176.eu-west-1.compute.internal  Started container
  Warning  Evicted                5m    kubelet, ip-192-168-66-176.eu-west-1.compute.internal  The node was low on resource: imagefs.
  Normal   Killing                5m    kubelet, ip-192-168-66-176.eu-west-1.compute.internal  Killing container with id docker://jnlp:Need to kill Pod
  Normal   Killing                5m    kubelet, ip-192-168-66-176.eu-west-1.compute.internal  Killing container with id docker://maven:Need to kill Pod

我该如何解决这个问题？我通常不完全理解 imagefs 是什么，我如何配置/增加它，或者避免饱和它。

附言。抱歉，这篇文章写得太被动了，我不得不使用积极的语气来使措辞足够冗长，这样我就可以不只是发布代码片段了。

Answer 1

解决;由于底层存储大小仅为 20gb，在 EBS 中更改为 50gb 并重新启动节点(增加了 nodefs)，从而消除了这个问题(因为 imagefs 不再饱和)。