[Gluster-devel] glusterfs file replication question
Krishna Srinivas
krishna at zresearch.com
Tue Feb 19 03:45:59 UTC 2008
Hi Roland,
The idea you want to use is correct, however afr does not handle
healing of holes
properly, we are working on it.
Regards
Krishna
On Feb 19, 2008 1:54 AM, <devzero at web.de> wrote:
> what if i combine AFR with Stripe Translator ?
>
> can this eventually help here?
>
> if i stripe a large file into several smaller parts and then open/write to that large file - can i combine this with AFR and will all the parts of that file be re-transmitted ?
>
> regards
> roland
>
>
> > -----Ursprüngliche Nachricht-----
> > Von: <devzero at web.de>
> > Gesendet: 17.02.08 22:23:14
> > An: "Krishna Srinivas" <krishna at zresearch.com>
>
> > CC: gluster-devel at nongnu.org
> > Betreff: Re: [Gluster-devel] glusterfs file replication question
>
>
> >
> > oh, that`s a real bandwidth killer / i/o hog then :)
> > think of 500gig disk with lot`s of large files , connected with 100mbit network.
> > if many of the files are open at the time of a node going down, that will take hours then for resync.
> >
> > is it planned to adress this to be handled more fficiently? for example just like fr1 from enbd ? (http://www.it.uc3m.es/ptb/nbd/#Intelligent_mirroring )
> >
> > regards
> > roland
> >
> > > -----Ursprüngliche Nachricht-----
> > > Von: "Krishna Srinivas" <krishna at zresearch.com>
> > > Gesendet: 17.02.08 18:14:21
> > > An: devzero at web.de
> > > CC: gluster-devel at nongnu.org
> > > Betreff: Re: [Gluster-devel] glusterfs file replication question
> >
> >
> > >
> > > Hi Roland,
> > > You are right. The entire file is copied.
> > > Krishna
> > >
> > > On Feb 17, 2008 5:12 PM, <devzero at web.de> wrote:
> > > > hello !
> > > >
> > > > first off, i think glusterfs is great. i have waited/searched very long for something like this!
> > > > thanks for making it !
> > > >
> > > > i`m new to glusterfs and have a question:
> > > >
> > > > if i use afr to have redundant copies of files to be stored on 2 or more nodes - when one node goes down and comes back online - what does this mean regarding transferred amount of data for "resync" ?
> > > >
> > > > say i had 10 files open at the time of the crash of a node, and each of those files is sized 1gb - will the "resync" then transfer 10gb of data over the network, i.e. doing a complete copy from the working node to the previously failed one ?
> > > >
> > > > regards
> > > > roland
> > > > sysadmin
> > > > _______________________________________________________________________
> > > > Jetzt neu! Schützen Sie Ihren PC mit McAfee und WEB.DE. 30 Tage
> > > > kostenlos testen. http://www.pc-sicherheit.web.de/startseite/?mc=022220
> > > >
> > > >
> > > >
> > > > _______________________________________________
> > > > Gluster-devel mailing list
> > > > Gluster-devel at nongnu.org
> > > > http://lists.nongnu.org/mailman/listinfo/gluster-devel
> > > >
> > >
> >
> >
>
>
> ________________________________________________________
> Bis 50 MB Dateianhänge? Kein Problem!
> http://www.digitaledienste.web.de/freemail/club/lp/?lp=7
>
>
>
>
> _______________________________________________
> Gluster-devel mailing list
> Gluster-devel at nongnu.org
> http://lists.nongnu.org/mailman/listinfo/gluster-devel
>
More information about the Gluster-devel
mailing list