File system fragmentation

Nifty Hat Mitch mitch48 at sbcglobal.net
Fri Oct 8 19:19:15 UTC 2004


On Fri, Oct 08, 2004 at 09:55:24AM -0400, Mark Haney wrote:
> On Fri, 2004-10-08 at 09:45, Douglas Furlong wrote:
> > On Fri, 2004-10-08 at 09:25 -0400, Mark Haney wrote:
> > > On Fri, 2004-10-08 at 09:14, Douglas Furlong wrote:
> > > > On Fri, 2004-10-08 at 09:08 -0400, Mark Haney wrote:
> > > > > What's the story with ext3 file fragmentation?  
> > > 

> > > > >From what I can tell of previous conversations, the fragmentation is
> > > > heavily dependant on free disk space.
....
> > > > You can use filefrag to work out how fragmented a file is, which is
> > > > handy for large files, no so handy for lots and lots of small files.
....
> I have plenty of space left on the partition (15GB or so out of 30GB)
> but I noticed that Evolution takes a while (a couple seconds ....

We do not know if it is file system IO that is slowing
your down, we can find out.

If you run "strace" on evolution with flags like -T -c -f -r we (linux
users) can measure the time spent in system calls like "read()" and
see if the time and associated byte count tell us anything.

In general I suspect other actions (not file fragmentation IO) are
taking more time.  But we do not know... it might be worth getting
some data to look at.

Absolutely compare the start up time of evolution the second time it is
started up with the first time.  Most files will be cached in DRAM so
the read() calls should be much quicker a second time.  There is no
fragmentation in file cache.  Look at the two traces side by side ....



-- 
	T o m  M i t c h e l l 
	Me, I would "Rather" Not.




More information about the fedora-list mailing list