lkml.org 
[lkml]   [1999]   [Feb]   [18]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: fsync on large files
"Stephen C. Tweedie" wrote:
> Linus, this makes very little sense right now: there is an _urgent_
> need for a journaled filesystem for Linux, and there is an urgent need
> for a solution which is of production quality. Implementing a new
> filesystem from scratch is hardly the way to achieve that. The
> recovery time of ext2 on large filesystems is a real showstopper to
> many potential applications.

As someone trying to actually deploy 80+ GB
RAIDed database servers right now on Linux, I can
certainly second Stephen's opinion. About the
*only* problem with ext2 right now is the
long fsck times. I've attached some measurements
to show you how serious the problem is.

Rich.

---- from posting to linux-raid
$ df
Filesystem 1024-blocks Used Available Capacity Mounted on
/dev/hda2 7802206 1127480 6270290 15% /
/dev/md0 85570400 56 81152504 0% /mnt/raid

# time mount /dev/md0 /mnt/raid
0.00user 0.25system 0:10.99elapsed 2%CPU (0avgtext+0avgdata
0maxresident)k
0inputs+0outputs (94major+12minor)pagefaults 0swaps

[ That's 11 seconds ]

To measure fsck performance, since it took about 2 hours
to actually *fill* the drive to 11% full, I had to resort
to extrapolation. Here are my hard results.

% Disk full Data blocks Time
---------------------------------------------
5% 3881184 5m35s (335s)
7% 6015252 7m33s (453s)
11% 8961576 9m59s (599s)
The data consisted of numerous copies of /usr/doc from
RedHat, therefore it's a fairly representative mix of
small files.

Extrapolating using an *extremely dodgy* linear
approximation gives the following estimates:

50% 40m (2400s)
90% 66m40s (4000s)

Rich.


--
Bibliotech Ltd., 631-633 Fulham Rd., London SW6 5UQ. +44 171 460 4646
http://www.bibliotech.co.uk/ ------ Home page: http://www.annexia.org/
Work email: rich@bibliotech.co.uk ----- Home:
rich@annexia.demon.co.uk
Original message content Copyright © 1999 Richard Jones
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:50    [from the cache]
©2003-2014 Jasper Spaans. hosted at Digital Ocean