IWETHEY v. 0.3.0 | TODO
1,095 registered users | 1 active user | 0 LpH | Statistics
Login | Create New User
IWETHEY Banner

Welcome to IWETHEY!

New Linux XFS max file count
I have a chunk of space on an XFS partition I'm using.
I'm throwing PDF page files into a single directory.
I'm about to hash them into buckets, create matching dirs, and link them around if it seems like a good idea.

I have 400,000 PDF files, and when I'm done I'd have 2.2 million.

I don't mind having LOTS of files if it won't make the dir too creaky.

I get them via a large unique id in the file name, but I have to glob to get all the pages for a single item.

Stupid?

I may have another partition to play with tomorrow, and at that point I can choose my file system. Resier used to rule on the many/small file issue. Or should I just use ext3 with dir indexing turned on?
New I don't think there's a hard limit.
http://oss.sgi.com/p...s_02_overview.pdf (12 page .PDF) doesn't mention hard limits for files per directory. Presumably any limit falls out of the other mentioned limits cited there.

As to whether it's a good idea, I dunno. It would scare me, but I know you've been doing things like this for ages.

Good luck!

Cheers,
Scott.
New For that many files, ext3 could be out of the game.
I haven't heard of directory indexing on ext3; it may make a difference. In years gone by, you wouldn't put even 40,000 files into an ext3 directory. :-)

Otherwise, XFS, JFS or Reiser (v3) are all viable for that many files.

Wade.

"Ah -- I take it the doorbell doesn't work?"
New I've been playing
XFS for now, and I like it.


100,000 files in the dir:

$ time find 000001_to_100000/ -name "pg_88298276500-??.pdf" -print
000001_to_100000/pg_88298276500-01.pdf
000001_to_100000/pg_88298276500-02.pdf
000001_to_100000/pg_88298276500-03.pdf
000001_to_100000/pg_88298276500-04.pdf
000001_to_100000/pg_88298276500-05.pdf
000001_to_100000/pg_88298276500-06.pdf
000001_to_100000/pg_88298276500-07.pdf
000001_to_100000/pg_88298276500-08.pdf

real 0m0.191s
user 0m0.110s
sys 0m0.080s


330,000 files (links to other dirs)

$ time find links/ -name "pg_88298276500-??.pdf" -print
links/pg_88298276500-01.pdf
links/pg_88298276500-02.pdf
links/pg_88298276500-03.pdf
links/pg_88298276500-04.pdf
links/pg_88298276500-05.pdf
links/pg_88298276500-06.pdf
links/pg_88298276500-07.pdf
links/pg_88298276500-08.pdf

real 0m0.600s
user 0m0.360s
sys 0m0.240s


If it goes to 5 seconds to whip through 2.2 million files, it'll for work me.
     Linux XFS max file count - (crazy) - (3)
         I don't think there's a hard limit. - (Another Scott)
         For that many files, ext3 could be out of the game. - (static) - (1)
             I've been playing - (crazy)

GPLed, for your pleasure.
36 ms