System Container Exits Unexpectedly

Symptom

After a training job is created, the system container exits unexpectedly.

Figure 1 Error logs

Possible Causes

The possible causes are as follows:

1. An error occurred in OBS.
   1. Unavailable file: The specified key does not exist.
   2. Insufficient OBS permissions
   3. OBS traffic limiting
   4. Others
2. The disk space is insufficient.

Solution

1. For an OBS error:
   1. Unavailable file: The specified key does not exist.

      For details, see Error Message "errorMessage:The specified key does not exist" Displayed in Logs.

   2. Insufficient OBS permissions

      For details, see What Should I Do If Error "stat:403 reason:Forbidden" Is Displayed in Logs When a Training Job Accesses OBS.

   3. OBS traffic limiting

      For details, see Error Message "BrokenPipeError: Broken pipe" Displayed When OBS Data Is Copied.

   4. Others

      For details, see OBS Server-Side Error Codes. Alternatively, obtain the request ID and contact OBS customer service.

2. For insufficient disk space:

   For details, see Common Issues Related to Insufficient Disk Space and Solutions.