[Pgcluster-general] large objects replication

[Koen Martens]

On Mon, Jul 16, 2007 at 05:18:09PM +0200, Benoit Mathieu wrote:
> OK.
> 
> If I import large objects from a shared directory, everything goes well, 
> and large objects are available from any cluster DB.
> 
> thanks !

Just curious, since i never got this to work:

What happens when you access the large object with lo_open, and make modifications with lo_write? Are those replicated?? I never got that to work.

Another problem I encountered was that if i pg_restore a db with large objects, replication of the large object fails (which is to be expected of course).

These limitations made me abandon pgcluster for that particular project. I think for static large objects, it works fine. But when you modify the large objects in stream access mode (lo_read / lo_write), things go awry.

Gr,

Koen

> 
> benoit mathieu
> 
> a.mitani at sra-europe.com a écrit :
> > Hi,
> >
> > PGCluster is replicating the file path of large object during each cluster
> > DB.
> > Therefore, the target file should to be able to access with same file path
> > from each cluster DB.
> > I recommend to use file server in order to store target large object
> > files, and share it during each cluster DB.
> >
> > Regards,
> > ------------
> > At.Mitani
> >
> >   
> >> Replication works when I create database, table, insert element.
> >> I've imported my file (file1) on cluster1 via the lo_import API of libpq
> >> C client library. It succeed, and I can export my file from cluster1,
> >> but I can't export my file from cluster2. I seems that large object
> >> hasn't been replicated ...
> >>
> >> On some old post in mailing-list I found that replication of large
> >> objects works since pgcluster 1.3, but only if files are stored on a
> >> shared filesystem. I can't find anything about setting a particular path
> >> to store large object in postgres. I've watched for differences in
> >> $PG_DATA between cluster1 and cluster2. The only difference is in
> >> directory $PG_DATA/pg_xlog
> >>
> >> Does someone have any idea on which directory to share between
> >> clusterDBs to manage large objects replication ? or any other advice to
> >> deal with this ?
> >>
> >> best regards,
> >>
> >> Benoit M.
> >>
> >>
> >> _______________________________________________
> >> Pgcluster-general mailing list
> >> Pgcluster-general at pgfoundry.org
> >> http://pgfoundry.org/mailman/listinfo/pgcluster-general
> >>
> >>     
> >
> > _______________________________________________
> > Pgcluster-general mailing list
> > Pgcluster-general at pgfoundry.org
> > http://pgfoundry.org/mailman/listinfo/pgcluster-general
> >
> >
> >   
> 
> _______________________________________________
> Pgcluster-general mailing list
> Pgcluster-general at pgfoundry.org
> http://pgfoundry.org/mailman/listinfo/pgcluster-general
> 
> 

-- 
K.F.J. Martens, Sonologic, http://www.sonologic.nl/
Networking, hosting, embedded systems, unix, artificial intelligence.
Public PGP key: http://www.metro.cx/pubkey-gmc.asc
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
Url : http://pgfoundry.org/pipermail/pgcluster-general/attachments/20070716/aece92ed/attachment.bin