Re: Null files reloaded :-)

[Ricardo Correia <wizeman@xxxxxxxx>]

On Tuesday 20 July 2004 19:06, Chris Wedgwood wrote:
> there is no way to recognise this, and there is no old file (it was
> unlinked and/or truncated)

Hmm.. I don't get it..

Consider your previous scenario:

1 - File A is safe on disk
2 - Process writes file B (metadata on disk, file data in RAM)
3 - Process renames B to A
4 - Periodic disk sync (every 5 seconds or so?)

Now if power failure occurs after 3, but before 4.. isn't the inode and 
content of A still on disk? And isn't the directory still pointing to the 
inode of A? I thought metadata at this point was only written to the journal.

If you could enlighten me, I would be appreciated :)
I just think it doesn't make much sense (but what do I know?) :-)