[olug] Anyone defragging their linux servers?

DYNATRON tech dynatron at gmail.com
Wed Oct 8 17:46:02 UTC 2008


the only situation i could think of would be if you installed on a FAT
partition. for ext3, on a low use web server, it's unheard of in my
experience. if the available tools to do such a thing are potentially
destructive, then i would say it's a no-brainer.

On 10/8/08, Will Langford <unfies at gmail.com> wrote:
>
> I recall defragging ext2 on 2.0 kernels ages ago.  Ya gotta boot single
> user
> mode or similar to try to check any normally-live fs... all the defraggers
> are 'offline'.  There's a few other user-land defraggers, but I think they
> work just by copying the files on top of themselves, hoping the low level
> drivers will defragment the new file as writing.  I think... 'shake' (???)
> was the name of the tool.  I would steer clear of e2fsdefrag.  You'll eat
> ext3 fs's with it.
> ext2/3 is very resistant to fragmentation, and for a web server that's not
> writing lots of decently sized files (ie: some kind of media streamer that
> is accepting input from one to possibly multiple sources), fragmentation
> shouldn't be a concern.
>
> You can always check the fragmentation level of a volume with `e2fsck -n
> /dev/foo`.  -n being 'dont do anything' ... and after it finishes all the
> tests, it'll print out a single non-contiguous line.
>
> From my understanding, NTFS also attempts to resist fragmentation.
>
> The short answer on ext3 defragging is that it's not going to happen.  It's
> been such a non-concern for the most part, that no tools exist to do it.  I
> find this decision rather lacking for the exact reason why you're asking -
> if the user feels better by doing it... and it's non detrimental to the FS,
> let him!  Which undoubtedly has management types of the opinion that ext3
> is
> therefor lacking a critical ability/tool.  Note that reiserfs also lacks a
> defragmenter (at least reiser3 does, dunno about 4 or newer).
>
> I believe ext4 is supposed to have some kind of defragmenting tool provided
> with it.  I'm not entirely sure why...
>
> -Will
> _______________________________________________
> OLUG mailing list
> OLUG at olug.org
> https://lists.olug.org/mailman/listinfo/olug
>



-- 
dynatron digital services
box 191 - 68037
www.dynatron.org
dynatron at gmail.com



More information about the OLUG mailing list