Re: FSSetForkSize writing massive amounts of data over network?
Re: FSSetForkSize writing massive amounts of data over network?
- Subject: Re: FSSetForkSize writing massive amounts of data over network?
- From: James Bucanek <email@hidden>
- Date: Wed, 2 Feb 2011 22:39:54 -0700
Mike <mailto:email@hidden> wrote (Wednesday,
February 2, 2011 5:14 PM -0800):
What are the contents of the data actually being written?
Zeros.
FSAllocateFork merely prepares an entry in the B-Tree and volume allocation
table. FSSetForkSize actually reserves all the blocks on the destination disk
that the file needs - which may be noncontiguous. For huge files, it may be
many millions of blocks all of which have to be noted in the allocation table
and if large enough, the extents overflow. Hence, such an operation requires
many (possibly millions) of small read and writes - all of which have to
happen over the network.
I suspect a heavily fragmented drive or a very full drive without a lot of
free space would show this behavior. Have you tried it with a completely empty
networked drive - a newly formatted large drive without anything on it?
No, it's the actual file/fill data. In one test, closing the
file after setting its EOF to 16GB saturated the network for 20
minutes--until I killed the process. The partially written file
contained nothing but zeros.
--
James Bucanek
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Filesystem-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden