Re: max size limit on posix xlator PROOF of CONCEPT patch

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

 



Thanks Avati

I need to re-estudy filesystem's basics, cause i just making erroneous asumptions about how fs and fuse actually works..

I see that trace xlator should be a good stating point for a quota translator as it intercepts every available syscall so its easy to
implement the requiered logic as most of the code is already in place.

The real target is to be able to trigger auto-pruning of files at certain mark provided from a quota xlator or simply combining both
functions to create a xlator suitable for local caching of files ( My autoprune proposed xlator). This is the first part. validating local existing
files is the other but afr should care of this i think. 

In the near term is nice for me be able to do minor modifications without burning my laptop or my neighbour's cat as i did code nothing in C
from the last 7 years!! :-P

Angel


El Miércoles, 16 de Enero de 2008 Anand Avati escribió:
> Angel,
>  Limiting filesystem size by tuning the params of statfs() is not possible.
> statfs values are only meant to be a reference to the user, and not limiting
> in any ways. So even if you get the tuning right, it will only be reflected
> in 'df' output, and the filesystem will actually still be able to grow than
> 1GB in size.
> 
>  What really needs to be done is filtering write() calls and other calls
> which create new inodes, This, in my opinion, is best done as a quota
> translator, with your requirements being a global quota option.
> 
> avati
> 
> 2008/1/17, Angel <clist@xxxxxx>:
> >
> > Hi, all
> >
> > This is my first proof of concept of limiting filsystem size on posix
> > xlator.
> > This far easy that implementing interception on filter module by now :-P
> >
> > test setup local posix xlator mounted on /home/sinosuke/mnt
> > mounted directory is void.
> >
> > cmd line is:
> >         glusterfs -N -L WARNING -l /dev/stdout -n trace-volume -f
> > /etc/glusterfs/client_posix.vol /home/sinosuke/mnt
> >
> > RESULTS
> >
> > Normal posix xlator (as 1.3.8tla363)
> > df -h
> > S.ficheros          Size Used  Avail Uso% Montado en
> > /dev/sda2             9,9G  4,6G  4,9G  49% /
> > udev                  502M   88K  502M   1% /dev
> > /dev/sda4              50G   21G   30G  42% /home
> > glusterfs             50G   21G   30G  42%  /home/sinosuke/mnt
> >
> > proof of concept posix xlator mountes /home/sinosuke/mnt with option
> > "max-size 1GB"
> >
> > sinosuke@sushiba:~/mnt> sinosuke@sushiba:~> df -h
> > S.ficheros          Size Used  Avail Uso% Montado en
> > /dev/sda2             9,9G  4,6G  4,9G  49% /
> > udev                  502M   88K  502M   1% /dev
> > /dev/sda4              50G   21G   30G  42% /home
> > glusterfs             1,0G  -29G   30G   -  /home/sinosuke/mnt
> >
> >
> > Other values ( used and vailable ) get insane values as they dont match
> > current filesize reported :-(
> >
> >
> > Comments welcome
> >
> >
> > Kind regards, Angel
> >
> >
> > --
> > ------------------------------------------------
> > Clist UAH
> > ------------------------------------------------
> >
> > _______________________________________________
> > Gluster-devel mailing list
> > Gluster-devel@xxxxxxxxxx
> > http://lists.nongnu.org/mailman/listinfo/gluster-devel
> >
> >
> >
> 
> 



-- 
------------------------------------------------
Clist UAH
------------------------------------------------




[Index of Archives]     [Gluster Users]     [Ceph Users]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Security]     [Bugtraq]     [Linux]     [Linux OMAP]     [Linux MIPS]     [eCos]     [Asterisk Internet PBX]     [Linux API]

  Powered by Linux