mirror of
https://github.com/Stone-Red-Code/dvc.git
synced 2026-11-11 15:28:27 +01:00
Merge pull request #178 from efiop/master
aws: use multipart to push changes
This commit is contained in:
+24
-2
@@ -7,6 +7,7 @@ import tempfile
|
||||
import requests
|
||||
import configparser
|
||||
import filecmp
|
||||
import math
|
||||
|
||||
from boto.s3.connection import S3Connection
|
||||
from google.cloud import storage as gc
|
||||
@@ -455,6 +456,28 @@ class DataCloudAWS(DataCloudBase):
|
||||
|
||||
return data_item
|
||||
|
||||
def _push_multipart(self, b, aws_key, fname):
|
||||
mp = b.initiate_multipart_upload(aws_key)
|
||||
|
||||
source_size = os.stat(fname).st_size
|
||||
chunk_size = 5000*1024*1024
|
||||
chunk_count = int(math.ceil(source_size / float(chunk_size)))
|
||||
|
||||
with open(fname, 'r') as fp:
|
||||
for i in range(chunk_count):
|
||||
offset = i * chunk_size
|
||||
left = source_size - offset
|
||||
size = min([chunk_size, left])
|
||||
part_num = i + 1
|
||||
|
||||
fp.seek(offset)
|
||||
mp.upload_part_from_file(fp=fp, part_num=part_num, size=size, cb=create_cb(fname))
|
||||
|
||||
if len(mp.get_all_parts()) != chunk_count:
|
||||
raise Exception("Couldn't upload all file parts")
|
||||
|
||||
mp.complete_upload()
|
||||
|
||||
def push(self, data_item):
|
||||
""" push, aws version """
|
||||
|
||||
@@ -473,8 +496,7 @@ class DataCloudAWS(DataCloudBase):
|
||||
key = bucket.new_key(aws_key)
|
||||
|
||||
try:
|
||||
key.set_contents_from_filename(data_item.resolved_cache.relative,
|
||||
cb=create_cb(data_item.resolved_cache.relative))
|
||||
self._push_multipart(bucket, aws_key, data_item.resolved_cache.relative)
|
||||
except Exception as exc:
|
||||
Logger.error('Failed to upload "{}": {}'.format(data_item.resolved_cache.relative, exc))
|
||||
return None
|
||||
|
||||
Reference in New Issue
Block a user