Re: [tlug] how to tune reiser4 for millions of files?

[Francois Cartegnie <fcartegnie@example.com>]

Le lundi 1 février 2010, Kalin KOZHUHAROV a écrit :
> Of course, nobody mentioned importing that into a normal database :-)
> MySQL will be happy to glob your data into a BLOB type, or you can
> save your self a good deal of time, if you parse the data from each
> file and input it in the database (iff the data is structured of
> course), then do:
No, i did :)
As he can't change the application's way of producing data (scattered files), 
i suggested writing them to a filesystem with database backend.
A "select mtime from inodes.." would return it's filenames candidates and he 
would have been free to access them. (fast if the read is done on a mirror 
which uses a regular fs).
Anyway, that's dirty hack, and indeed the application itself needs to be  
rewritten to use a regular db :/

This problem enlightens the gap between the filesystem's single index and the 
Desktop meta indexing/search engines (like Beagle).
Attributes indexing uses ? example: A file system with multiple indexes would 
help instantly detecting unauthorized changes where current security scripts 
are just scheduled and need to walk the entire filesystem.

Francois