[OpenAFS-devel] Adding an "estimate" function to vlserver

Neulinger, Nathan nneul@umr.edu
Fri, 4 Apr 2003 11:11:45 -0600


Definately not "amanda specific". My current backup system just uses the
vos examine size + 5% or so depending on the volume size class.
(small/medium/huge) If we added 'vos size', I would definately switch to
using it instead.

Problem is that vos examine is showing disk usage by quota. That's
completely different than how much space a dump will take.

For example, if you have a directory structure 30 levels deep filled
with tiny files, there will be a LOT of dump content overhead, but very
little quota usage. Opposite for a volume with a small number of LARGE
files.=20

In order to fix vos examine, you'd have to have it track quota based on
more detail of what the layout of the volume was. Doing it the way that
vos estimate/vos size would do would be way too slow to have 'vos
examine' do it. ve is usually fairly fast.=20

Additionally, there is no way in vos examine to get information in any
reasonable way about how much space an incremental dump will take. With
vos size actually looking at volume contents, that should be relatively
trivial to do. One question would be whether we could modify dumpstuff.c
to re-use existing dump code, but have it just keep track instead of
actually writing out the dump. Then again, might be simple enough to
just duplicate it.

-- Nathan

------------------------------------------------------------
Nathan Neulinger                       EMail:  nneul@umr.edu
University of Missouri - Rolla         Phone: (573) 341-4841
Computing Services                       Fax: (573) 341-4216


> -----Original Message-----
> From: Jim Rees [mailto:rees@umich.edu]=20
> Sent: Friday, April 04, 2003 11:05 AM
> To: openafs-devel@openafs.org
> Subject: Re: [OpenAFS-devel] Adding an "estimate" function to=20
> vlserver=20
>=20
>=20
> Except that this feature is not specific to Amanda.  I can=20
> see it being
> useful for a variety of reasons.
>=20
> One thing I'm a bit confused about is why we don't fix "vos=20
> exam" to give
> the more accurate estimate.  But I know almost nothing about=20
> the vldb so I'm
> sure there's a good reason.
> _______________________________________________
> OpenAFS-devel mailing list
> OpenAFS-devel@openafs.org
> https://lists.openafs.org/mailman/listinfo/openafs-devel
>=20