[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Size of mirror - different than said in "mirror size" page (bigger)

(good question so CC the list)

On Tue, Aug 18, 2009 at 12:24:00PM -0400, Hugo Segovia wrote:
> I'm setting up a private Debian mirror, and I'm using "ftpsync".

As a general reminder, don't forget that a proxy is often more adequate
for private uses (approx, apt-proxy..).

> In ftpsync.conf, I set the ARCH_EXCLUDE var like this:
> ARCH_EXCLUDE="alpha amd64 arm armel hppa hurd-i386 ia64 kfreebsd-amd64
> kfreebsd-i386 m68k mips mipsel powerpc s390 sh sparc"
> So, I think I'm mirroring only "i386", "source" and "all".

That's what happens.

> According to the "Mirror Size" page, the combined size of "i386",
> "source" and "all" is just about 130 GB.
> However, now I have almost 150 GB used by my mirror archive copy.

I cannot explain such a big difference (but make sure you are running
the last ftpsync from git with uptodate exclusion masks).
(And a fast check on a mirror that the figures displays kind of match
with the reality).

Which is your upstream mirror (maybe it forgot to remove some old
deprecated files ?)
When and how did you measure the 150 GB ?

* When: because during the sync itself, rsync (through ftpsync) will
keep the previous version of all files until all the new ones are ready
to be "moved in production". This temporally "buffer" use some disk

* How: disk usage is not the same as file size (depends on your fs, some
files are symlinks), 1024 vs 1000 multiples.

> So, why would that happen? Maybe I didn't set ARCH_EXCLUDE well? I want
> to mirror only the "i386" arch.

Actually, this page only shows the size of files contained in the pool,
from dak statistics: http://ftp-master.debian.org/~joerg/arch-space

Files describing the archive itself and providing debian-installer are
not taken into account, that makes around 1.3GB for i386.

In any case, a margin must be taken into account since the archive size
always change with time.


Simon Paillard

Reply to: