Log Overview

Read the corresponding section based on your product model.

Viewing Logs (Ascend EP) for the following product models:
- Atlas 200I/500 A2 inference products
- Atlas inference products
- Atlas training products
- Atlas A2 training products/Atlas A2 inference products
- Atlas A3 training products/Atlas A3 inference products
- Ascend 950PR/Ascend 950DT
Viewing Logs (Ascend RC) for Atlas 200I/500 A2 inference products only.
Viewing Logs (Open Control CPU Form) for Atlas inference products only.

Log Types

Logs record the running process and exception information of a system and support troubleshooting in system running and program debugging during development. Logs are classified into the following types:

System logs that record OS events on the device, including:
- System logs on the control CPU (kernel-mode logs and user-mode logs generated during system process running), which reflect the overall running status of the AI processor.
- System logs on the non-control CPU (for example, low-power component), which reflect the running status of components such as the low-power components, Task Scheduler, and ISP.
App logs generated during AI app runtime, including:
- Logs printed by components such as AscendCL, GE, Runtime, and HCCL on the host.
- Logs printed by the AI CPU process on the device.

Replace id and pid in log file names in this document with the actual device ID and service process ID, respectively.

During program running, the maintenance and test information of the software stack is recorded in the memory. If a program runtime error occurs or a process ends, the information is flushed to a file, preventing frequent generation and recording of log files from affecting the performance. The information is stored in trace logs, which are generated independently. For details, see Viewing Trace Logs.

Log Format

A log example is as follows:

[ERROR] TEFUSION(12940,atc):2021-10-17-05:54:07.599.074 [tensor_engine/te_fusion/pywrapper.cc:33]InitPyLogger Failed to import te.platform.log_util

The log format is as follows:

[Level] ModuleName(PID,PName):DateTimeMS [FileName:LineNumber]LogContent

**Table 1** Log field description
Field	Description
Level	Log level Options: ERROR, WARNING, INFO, and DEBUG.
ModuleName	Name of the module that generates a log
PID	Module process ID
PName	Module process name
DateTimeMS	Time when a log is printed Format: yyyy-mm-dd-hh:mm:ss.fff.zzz (year-month-day-hour:minute:second:millisecond:microsecond)
FileName:LineNumber	File name and the corresponding line number that calls the log printing API
LogContent	Detailed log content of each module