Documentation

Overview

ascend-log-collect.sh is used to collect operating environment information, Ascend NPU health information, Ascend software logs, system-level device logs, and MindX Logs during fault analysis and locating. The collected data is saved in tar.gz format. Decompress the collected tar.gz log package to obtain system-report.log.gz, ascend-report.log.gz, ascend_user_log.tar.gz (collected if it exists), ascend_user_install_log.tar.gz, and device_log.tar.gz. If device logs exist, the YYYY-MM-DD-HH-MM-SS folder is generated. If edge device logs exist, alog.tar.gz and plog.tar.gz are also obtained. If the path of the SDK logs to be collected is specified, mindx_sdk_info_*.tar.gz is obtained as well.

The logs collected may contain system information. Pay attention to the risk of information disclosure during the use of the exported logs.

  • The operating environment information (system-report.log.gz) includes:
    • Operating system information, which is obtained by reading the /etc/lsb-release and /etc/os-release files.
    • PCIe device information, which can be obtained by running the lspci command.
    • System software package information, which can be obtained by running the apt list/rpm -qa command.
    • Python software package information, which can be obtained by running the pip list/pip3 list command.
    • Firmware version and system version.

      Query the firmware version:

      /usr/local/Ascend/driver/tools/upgrade-tool --device_index -1 --component -1 --version

      Query the system version:

      /usr/local/Ascend/driver/tools/upgrade-tool --device_index -1 --system_version

      In the preceding command, /usr/local/Ascend indicates the driver installation path, which can be obtained from the /etc/ascend_install.info file.

    • Environment variable information, which can be obtained by running the env command.
    • Permission on the startup file, which can be obtained by running the ls -l /boot command.
    • Memory status, which can be obtained by running the cat /proc/meminfo command.
    • CPU status, which can be obtained by running the cat /proc/cpuinfo command.
    • Interruption report file, which can be obtained by running the cat /proc/interrupts command.
    • Slot information, which can be obtained by running the dmidecode -t slot command.
    • Last startup time of the system, which can be obtained by running the last reboot command.
  • The Ascend NPU health information (ascend-report.log.gz) includes:
    • Ascend software and hardware version information and Ascend software installation information.
    • Data such as Ascend-DMI version information, topology detection result, and device health status, which can be obtained by running the ascend-dmi command. For details, see Using the Ascend-DMI Tool.
    • Processor information, which can be obtained by running the npu-smi command.
    • System-level logs of the host
      • Host driver logs:

        Logs whose names contain ascend in the /var/log/syslog* directory.

      • Host kernel mode logs:

        Logs whose names contain ascend in the /var/log/messages* directory.

        Logs whose names contain ascend in the /var/log/kern.log* directory.

        Logs whose names contain ascend in the /var/log/kernel.log* directory

        Logs whose names contain ascend in the /var/log/dmesg* directory

  • ascend_user_log.tar.gz: Ascend software logs, mainly CANN application logs generated on the host and device.
    • Logs generated by applications on the host are stored in the ${HOME}/ascend/log/plog directory.
    • Logs generated by applications on the device are stored in the ${HOME}/ascend/log/device-<id> directory.
  • ascend_user_install_log.tar.gz: CANN installation logs. The installation logs of the root user are stored in /var/log/ascend_seclog, and the installation logs of non-root users are stored in ${HOME}/var/log/ascend_seclog.
  • device_log.tar.gz: collected Ascend software and hardware logs. The involved directory is /var/log/npu.
  • YYYY-MM-DD-HH-MM-SS folder: stores system-level logs of the device. Device logs are packaged into the log package in the current directory of the host.
  • alog.tar.gz: logs of edge device middleware (such as MindX Edge)
  • plog.tar.gz: O&M logs of edge devices
  • The collected SDK information (mindx_sdk_info_*.tar.gz) includes:

    Processor information (version and logs), OS version, environment variables, network information, and MindX SDK information (version, configuration file, logs, and third-party library versions)