Hi
In Linux file system Inodes, the generation no is
time stamp based, means when the first time the inode
is used after formatting volume, time stamping is used
and all the next reuses, increment the same, please
correct me if I am wrong.
Now the similar thing in NTFS is FileId consisting
of 8 byte no, 2 byts for sequence no + 6 bytes for MFT
index. Now my question is, how NTFS File Id gurantees
to be unique across different invocations of
formatting.
Consider the case. I have a fresh NTFS vol. and I
create 1.txt which gets assigned MFT entry 0x1B, and
sequence no 1. Now I delete it, again create it, the
same index is reused with sequence no. 2 now. This is
as expected.
Now I format volume. Again create the first file
1.txt and I can see, using NTFS disk explorer tool
that the new entry is using Index no 0x1B, with
sequence no 1 again.
How can I get gurantee of the uniquness of the
file Id after formatting as it is in Linux because of
use of time stamp in Inode generation no. This case is
considered for a NFS server on Windows, which
generates File Handles based on File Id, but if this
is the case(say only single file created in a exported
folder) and if some client has mounted the share and
using the same file, then this client has the file
handle(file id) for this file. Now I format the volume
on server and create the same different file with
same set up again. This file will get same MFT index
no, and same sequence no ie. 1, and the client side
handle will perfectly map to this file, if client
tries to access the share,which ideally should not
happen because this is different file indeed.
I experimented with the volume creation and all
this stuff. Didnt use NFS server and client but all
other
things I did at volume level and saw this problem.
So how do I solve this, any pointers. Please guide.
Thanks
Mrunal
Yahoo! India Matrimony: Find your partner now. Go to http://yahoo.shaadi.com