Upgrading Images

This section explains how to upgrade the binary file within a container image of the same version. The upgrade does not alter permissions or startup parameters. For details about this upgrade mode, see Upgrade Description.

Upgrading Management Node Components

  1. Create an image using the new software package by referring to Preparing an Image.

    Ensure that the image names are the same. Otherwise, the original configuration file may fail to start the pod.

  2. Query the Deployment configuration of the old version.
    kubectl get deployment -A|grep {Component name}

    Command output (ClusterD as an example):

    mindx-dl         clusterd        1/1     1      1       45h
  3. Restart Deployment.
    kubectl rollout restart deployment -n {Namespace name} {Deployment name}

    Command output (ClusterD as an example):

    deployment.apps/clusterd restarted
  4. Check whether the pod of the new version is started.
    kubectl get pod -A|grep {Component name}

    (ClusterD as an example) If the command output is as follows, the pod is started successfully.

    mindx-dl   clusterd-99f8795c8-drqb4  1/1  Running 0       1m

Upgrading Compute Node Components

  1. Create an image using the new software package by referring to Preparing an Image.

    Ensure that the image names are the same. Otherwise, the original configuration file may fail to start the pod.

  2. Query the DaemonSet configuration of the old version.
    kubectl get ds -A|grep {Component name}

    Command output (NodeD as an example):

    mindx-dl         noded        1/1     1      1       45h
  3. Restart DaemonSet.
    kubectl rollout restart ds -n {Namespace name} {DaemonSet name}

    Command output (NodeD as an example):

    daemonsets.apps/noded restarted
  4. Check whether the pod of the new version is started.
    kubectl get pod -A|grep {Component name}

    (NodeD as an example) If the command output is as follows, the pod is started successfully.

    mindx-dl   noded- m4j4r  1/1  Running 0     1m