openstack
/
nova
Code Issues Proposed changes

VMware: specify chunk size when reading image data 

When reading OVA image data from Glance without specifying chunk size,
nova-compute tries to read the entire tar stream in memory because this
is how the tar fileobject is implemented. This patch fixes this by
specifying the chunk size to be 64KB.

Change-Id: I953dfa1c28c25886306e1c9afe281cf2faca4db5
Closes-Bug: #1495834
(cherry picked from commit 71d2d31b9b)
This commit is contained in:
Radoslav Gerganov 2015-10-06 16:01:45 +03:00 committed by Dongcan Ye
parent 8e741e3b25
commit 51bfbbd829
1 changed files with 2 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
nova/virt/vmwareapi/io_util.py
View File

@ -34,6 +34,7 @@ IMAGE_API = image.API()
IO_THREAD_SLEEP_TIME = .01
GLANCE_POLL_INTERVAL = 5
CHUNK_SIZE = 64 * 1024 # default chunk size for image transfer
class ThreadSafePipe(queue.LightQueue):
@ -173,7 +174,7 @@ class IOThread(object):
self._running = True
while self._running:
try:
data = self.input.read(None)
data = self.input.read(CHUNK_SIZE)
if not data:
self.stop()
self.done.send(True)