Detailed Description of Data Sources

Overview

This document provides a detailed description of the various data sources supported by Ascend-faultdiag-toolkit, including:

• Data types and specific data items
• Data collection methods (SSH online collection/offline log parsing)
• Data source locations (specific commands/file paths)

The Ascend-faultdiag-toolkit supports data source collection from the following three device types:

• Host
• Baseboard Management Controller (BMC)
• Switch

Each device supports the following two collection methods:

• SSH online collection: Connects to the device via SSH and executes commands to obtain real-time data.
• Offline log parsing: Parses exported log files to extract historical data.

Host Data Source

SSH Online Collection

Offline Log Parsing

Host offline logs support three different configuration versions, each with its own set of file paths and parsing keywords.

Version 1 Configuration (ParseConfigCollectionV1)

File Structure

log_directory/
├── hccn_tool.log        # Network Configuration Tool Log
├── npu_card_info.log    # NPU Card Information Log
├── pcie_info.log        # PCIe Information Log
└── version_info.log     # Version Information Log

Data Items and Parsing Configuration

Version 2 Configuration (ParseConfigCollectionV2)

File Structure

log_directory/
├── hccn_log/
│   ├── net_conf.log     # Network Configuration Log
│   ├── optical.log      # Optical Module Log
│   └── stat.log         # Statistics Log
├── npu_smi_log/
│   └── npu_smi.log      # NPU SMI Log
└── pcie_log/
    └── pcie.log         # PCIe Log

Data Items and Parsing Configuration

Version 3 Configuration (ParseConfigCollectionV3)

File Structure

log_directory/
├── lldp.log        # LLDP Log
└── optical.log     # Optical Module Log

Data Items and Parsing Configuration

MSNPUREPORT Logs

In addition to the above configuration, parsing of logs generated by the MSNPUREPORT tool is also supported:

• Path: Timestamp-named subdirectory under the log directory (e.g., 2023-10-01_14-30-00)
• Content: Detailed NPU information

BMC Data Source

SSH Online Collection

Data Items and Sources

Diagnostic Information Compressed Package

• Remote path: /tmp/dump_info.tar.gz
• Local storage path: CommonPath.TOOL_HOME_BMC_DUMP_CACHE_DIR
• Naming format: {host}_{sn_num}_{subfix_date_time}.tar.gz

Offline Log Parsing

Log Source

BMC offline logs are typically obtained through the diagnostic information archive (dump_info.tar.gz) generated by the ipmcget -d diaginfo command. After decompression, the directory structure is as follows:

dump_info/
└── AppDump/
    ├── bmc_network/
    │   └── network_info.txt      # Network information file
    ├── frudata/
    │   └── fruinfo.txt           # FRU information file
    ├── event/
    │   ├── sel.txt               # System event log
    │   └── current_event.txt     # Current health event
    ├── sensor/
    │   └── sensor_info.txt       # Sensor information
    ├── network_adapter/
    │   └── optical_module/
    │       └── optical_module_history_info_log.csv  # Optical module history log 1
    └── CpuMem/
        └── NpuIO/
            └── optical_module_history_info_log.csv  # Optical module history log 2

Data Items and Sources

Switch Data Source

SSH Online Collection

Offline Log Parsing

Switch offline logs primarily come in the following two forms:

• CLI command output log: Contains the execution results of various switch commands.
• Diagnostic information output:Structured logs generated by switch diagnostic tools.

Log File Locations

• CLI output log: Usually a single file (such as switch_cli_output.txt) or multiple files categorized by command.
• Diagnostic information output: Usually a file with a diagnostic marker (such as diag_info.txt).

Data Types and Sources