AWS S3 多部分上传时，如果在上传过程中出现错误或中断，就需要使用断点续传来继续上传。但是有时候会出现无法恢复上传的情况，如上传时断电或者程序异常退出等导致已上传部分数据丢失，就无法继续上传。这种情况下需要重新开始上传。以下是一个示例，演示如何使用AWS SDK进行多部分上传的实现：

import boto3
s3 = boto3.resource('s3')
bucket = s3.Bucket('my-bucket-name')
# Set the desired multipart chunk size
chunk_size = 8 * 1024 * 1024
# Create a new multipart upload and obtain the upload id
mpu = bucket.initiate_multipart_upload(Key='my-large-file.mp4')
upload_id = mpu.id
# Open the file for reading in binary mode
file = open('my-large-file.mp4', 'rb')
try:
    # Read the first chunk
    data = file.read(chunk_size)
    part_number = 1
    # Loop over the chunks and upload them
    while data:
        # Create a new part and upload it
        part = mpu.Part(part_number)
        response = part.upload(Body=data)
        # Print the response
        print('Part #{0} uploaded: {1}'.format(part_number, response['ETag']))
        # Read the next chunk
        data = file.read(chunk_size)
        part_number += 1
except Exception as e:
    # Abort the upload on error
    mpu.abort()
    print('Error occurred during upload: {0}'.format(e))
else:
    # Complete the upload when done
    mpu.complete()
    print('Upload completed successfully.')
finally:
    # Close the file handle
    file.close()

要实现断点续传，需要记录已上传分块的 ETag 和 part number。当出现错误时，可以通过调用 bucket.MultipartUpload('my-large-file.mp4').abort()来终止当前上传任务并删除已上传