昇腾社区首页
中文
注册
开发者
下载
支持
积分兑换 NEW

升级Container Manager

在物理机上直接替换Container Manager二进制升级组件。

  1. 以root用户登录Container Manager组件部署的节点。
  2. 将获取到的Container Manager软件包上传至服务器的任意目录（如“/tmp/container-manager”）。
  3. 进入“/tmp/container-manager”目录并进行解压操作。
    unzip Ascend-mindxdl-container-manager_{version}_linux-{arch}.zip

    <version>为软件包的版本号；<arch>为CPU架构。

  4. 依次执行以下命令，升级Container Manager组件。
    # 停止Container Manager系统服务，并删除对应Container Manager二进制文件
systemctl stop container-manager.service
chattr -i /usr/local/bin/container-manager
rm -f /usr/local/bin/container-manager

# 从解压文件中获取新二进制文件，替换旧Container Manager二进制文件
cp /tmp/container-manager/container-manager /usr/local/bin
chmod 500 /usr/local/bin/container-manager

# 重启Container Manager系统服务
systemctl daemon-reload
systemctl start container-manager.service
  5. 验证Container Manager组件的升级状态。
    1. 查看组件服务的状态，状态应为active (running)。
      systemctl status container-manager.service

      回显示例：

       1
 2
 3
 4
 5
 6
 7
 8
 9
10
      ● container-manager.service - Ascend container manager
     Loaded: loaded (/etc/systemd/system/container-manager.service; disabled; vendor preset: enabled)
     Active: active (running) since Wed 2025-11-26 20:56:50 UTC; 16s ago
    Process: 41459 ExecStart=/bin/bash -c container-manager run  -ctrStrategy ringRecover -logPath=/var/log/mindx-dl/container-manager/container-manager.log >/dev/null 2>&1 & (code=exited, status=0/SUCCESS)
   Main PID: 41464 (container-manag)
      Tasks: 10 (limit: 629145)
     Memory: 13.3M
     CGroup: /system.slice/container-manager.service
             └─41464 /home/container-manager/container-manager run -ctrStrategy ringRecover
...
    2. 查看组件日志。
      cat /var/log/mindx-dl/container-manager/container-manager.log

      回显以Atlas 800I A3 超节点服务器为例：

      1
2
3
4
5
6
7
      [INFO]     2025/11/25 22:46:59.007163 1       hwlog/api.go:108    container-manager.log's logger init success
[INFO]     2025/11/25 22:46:59.007288 1       command/run.go:150    init log success
[INFO]     2025/11/25 22:46:59.007506 1       devmanager/devmanager.go:134    get card list from dcmi reset timeout is 60
[INFO]     2025/11/25 22:46:59.250103 1       devmanager/devmanager.go:142    deviceManager get cardList is [0 1 2 3 4 5 6 7], cardList length equal to cardNum: 8
[INFO]     2025/11/25 22:46:59.250267 1       devmanager/devmanager.go:171    the dcmi version is 25.5.0.b030
[INFO]     2025/11/25 22:46:59.250405 1       devmanager/devmanager.go:235    chipName: Ascend910, devType: Ascend910A3
...

      如果出现如下打印信息，表示组件运行正常。

      ...
[INFO]     2025/11/25 22:46:59.289352 1       devmgr/workflow.go:57    init module <hwDev manager> success
[INFO]     2025/11/25 22:46:59.293773 1       app/config.go:40    load fault config from /home/faultCode.json success
[INFO]     2025/11/25 22:46:59.293866 1       app/workflow.go:50    init module <fault manager> success
[INFO]     2025/11/25 22:46:59.293901 1       app/workflow.go:76    init module <container controller> success
[INFO]     2025/11/25 22:46:59.293930 1       app/workflow.go:64    init module <reset-manager> success
[INFO]     2025/11/25 22:46:59.315101 378     devmgr/hwdevmgr.go:365    subscribe device fault event success
...
父主题： 升级