30M 左右的 parquet 文件， 转成 pandas 的 dataframe 存 mysql， 直接耗完 6 个 g 的内存?

[scalaer]

python 代码

from pyarrow import parquet as pq


fs = s3fs.S3FileSystem(...)

path = ['s3a://.....']

r = pq.ParquetDataset(_path, filesystem=fs)
df = r.read().to_pandas()
df.to_sql(....)

---

相关的日志

Jan 18 14:18:33  kernel: Out of memory: Kill process 3342 (python3) score 842 or sacrifice child
Jan 18 14:18:33  kernel: Killed process 3342 (python3) total-vm:7950428kB, anon-rss:6945048kB, file-rss:0kB, shmem-rss:0kB

[scalaer]

漏了点处理
df['_time'] = df['time'].apply(lambda x: parse(x[:11] + " " + x[12:]).strftime("%Y-%m-%d %H:%M:%S"))
df['bytes'] = df['request_size'] + df['response_size']
df['bytes'] = df['bytes'].astype(int)