[Nepomuk] Backup service. Was: Re: ontology related advice request

Evgeny Egorochkin phreedom.stdin at gmail.com
Mon Oct 26 12:52:45 CET 2009


В сообщении от Понедельник 26 октября 2009 12:44:39 автор Sebastian Trüg 
написал:
> Just for your consideration:
> 
> sopranocmd --backend sesame2 --serialization nquads --nrl --graphselect
> "s;p;o;g" query 'select distinct ?s ?p ?o ?g where { graph ?g { ?s ?p ?o .
>  } . OPTIONAL { { ?g nie:dataGraphFor ?x . } UNION { ?g
>  nrl:coreGraphMetadataFor ?g2 . ?g2 nie:dataGraphFor ?x . } UNION { ?g
> <http://www.strigi.org/fields#indexGraphFor> ?x . } UNION { ?g
> nrl:coreGraphMetadataFor ?g2 . ?g2
> <http://www.strigi.org/fields#indexGraphFor> ?x . } } . FILTER(!BOUND(?x))
>  . FILTER(?g != <urn:crappyinference:inferredtriples>) . }'
> 
> This is the weird command that only exports statements that are not created
>  by strigi or akonadi and are not inferred by the crappy inferencer.
> 
> Maybe I should blog this...

I knew a cli command like this existed. This however doesn't address other 
backup tasks like backing up different kinds of data to different locations, 
proper labelling, encryption and what not. Duplicity comes to mind as a 
model...

> On Friday 23 October 2009 07:47:12 Evgeny Egorochkin wrote:
> > В сообщении от Пятница 23 октября 2009 08:18:15 автор Ben Martin написал:
> > > On Wed, 2009-10-21 at 17:04 +0100, Laura Dragan wrote:
> > > ...
> > >
> > > > I promised a while back that I would do a backup service
> > > > (NepomukService) but I never got around to do it. There are a lot of
> > > > things to consider, especially regarding the decision on what to
> > > > backup .. We should discuss this at the workshop, maybe we can
> > > > actually get it done :)
> > >
> > > I've also been thinking it would be nice to make such a thing more
> > > readily available. Currently I just sopranocmd export to a file which
> > > is itself incrementally backed up by bacula.
> >
> > For now it seems like the diff tool should have some presets for
> >  backup/restore operations. eg you want your work data backed up daily
> > and to a different location than tags for your multimedia lib which is
> > backed up weekly.
> >
> > Looks like a collection of small scripts + some GUI. I wish I had time to
> >  do this but unf I'm totally occupied with libstreamanalyzer atm :(
> >
> > > But having some sort of rdiff
> > > like thing build into the backup process would make sense.
> > >
> > > For redland, a scheduled repair and lzo db_dump -p of the db files
> > > seems to work OK.
> >
> > It's better to only backup user-generated content. This means you get to
> > backup MBs instead of GBs. Imagine someone telling nepomuk to index KDE
> >  repo or something equally crazy(but possible) like this.
> >
> > Of course indexing GBs of data doesn't apply to redland :)
> >
> > -- Evgeny


More information about the Nepomuk mailing list