Hi Rupert,

tried again on one of the machines and I still get the same behaviour.

What actually happens is:
1) I start Stanbol and the default dbpedia index is installed in
dbpedia-2013.01.15

This dir is about 100MB

2) I add the dbpedia.solrindex.zip (which is about 3.5 GB) to
stanbol/datafiles and at that moment two different folders are created
dbpedia-2013.01.15-1
dbpedia-2013.01.15-2

They grow to the same size (about 9.5 GB)

After the process stops dbpedia-2013.01.15 is deleted (the default index).

The ref file points to dbpedia-2013.01.15-2

I have checked the logs, but can only see references to an error which
states "Too many close" on the SolrCore:

15.01.2013 15:16:14.493 *INFO* [Thread-44]
org.apache.stanbol.commons.solr.managed.impl.ManagedSolrServerImpl  ...
start to ACTIVATE Index dbpedia on ManagedSolrServer
15.01.2013 15:16:19.493 *INFO* [DataFileTrackingDaemon]
org.apache.stanbol.commons.stanboltools.datafileprovider.impl.tracking.DataFileTrackerImpl
 ... tracking stopped!
15.01.2013 15:22:57.477 *WARN* [OsgiInstallerImpl]
org.apache.solr.handler.component.SpellCheckComponent No queryConverter
defined, using default converter
15.01.2013 15:22:57.526 *INFO* [OsgiInstallerImpl]
org.apache.stanbol.commons.solr.RegisteredSolrServerTracker  ... in
addingService for IndexReference[server:null,index:dbpedia] (ref:
[org.apache.solr.core.SolrCore])
15.01.2013 15:22:57.529 *INFO* [OsgiInstallerImpl]
org.apache.stanbol.commons.solr.managed Service [173] ServiceEvent
REGISTERED
15.01.2013 15:22:57.530 *INFO* [OsgiInstallerImpl]
org.apache.stanbol.commons.solr.RegisteredSolrServerTracker  ... in
removedService for IndexReference[server:null,index:dbpedia] (ref:
[org.apache.solr.core.SolrCore], service
org.apache.solr.client.solrj.embedded.EmbeddedSolrServer@5dcfd078)
15.01.2013 15:22:57.606 *INFO* [OsgiInstallerImpl]
org.apache.stanbol.commons.solr.managed Service [167] ServiceEvent
UNREGISTERING
15.01.2013 15:22:57.616 *ERROR* [OsgiInstallerImpl]
org.apache.solr.core.SolrCore Too many close [count:-1] on
org.apache.solr.core.SolrCore@33318b82. Please report this exception to
solr-u...@lucene.apache.org
15.01.2013 15:22:57.644 *WARN* [Thread-44]
org.apache.solr.handler.component.SpellCheckComponent No queryConverter
defined, using default converter
15.01.2013 15:22:57.693 *INFO* [Thread-44]
org.apache.stanbol.commons.solr.RegisteredSolrServerTracker  ... in
addingService for IndexReference[server:null,index:dbpedia] (ref:
[org.apache.solr.core.SolrCore])
15.01.2013 15:22:57.697 *INFO* [Thread-44]
org.apache.stanbol.commons.solr.managed Service [174] ServiceEvent
REGISTERED
15.01.2013 15:22:57.698 *INFO* [Thread-44]
org.apache.stanbol.commons.solr.RegisteredSolrServerTracker  ... in
removedService for IndexReference[server:null,index:dbpedia] (ref:
[org.apache.solr.core.SolrCore], service
org.apache.solr.client.solrj.embedded.EmbeddedSolrServer@f50dbe6)
15.01.2013 15:22:57.700 *INFO* [Thread-44]
org.apache.stanbol.commons.solr.managed Service [173] ServiceEvent
UNREGISTERING
15.01.2013 15:22:59.307 *ERROR* [Thread-44] org.apache.solr.core.SolrCore
Too many close [count:-1] on org.apache.solr.core.SolrCore@4c458d2b. Please
report this exception to solr-u...@lucene.apache.org
15.01.2013 15:22:59.311 *INFO* [Thread-44]
org.apache.stanbol.commons.solr.managed.impl.ManagedSolrServerImpl  ...
Index dbpedia on ManagedSolrServer default is now ACTIVE

It is like the installing phase starts with two different concurrent
threads and ends up with two different indexes (which are actually the
same).

What do you think the cause could be?

Regards
Andrea

2013/1/15 Rupert Westenthaler <rupert.westentha...@gmail.com>

> On Tue, Jan 15, 2013 at 3:48 PM, Andrea Di Menna <ninn...@gmail.com>
> wrote:
> > a) dbpedia-2013.15.1
>
> This is the folder created for the DBpedia default index that is
> included in the Stanbol launcher
>
> > b) dbpedia-2013.15.1-1
>
> This is the folder created for the dbpedia.solrindex.zip in the datafiles
> folder
>
> (a) should get deleted as soon as (b) is fully copied, initialized and
> added as SolrCore to the CoreContainer.
>
>
> >
> > when the zip file has been uncompressed both folders are the same size
> and
> > contain the same data, and they are both kept on the HDD.
>
> (a) should be ~100MByte in size. (b) depends on the
> dbpedia.solrindex.zip you are using. (a) and (b) should not be the
> same data.
>
> >
> > Of course, in a way this makes the index double its size.
> >
> > I remember seeing this pattern applied also when deploying Stanbol in a
> > Jetty instance, but as far as I remember one of the two folders was
> delete
> > when the installation process ended.
>
> Exactly this is the expected behavior
>
> > What could be the cause of this behaviour?
> >
>
> I have never seen this happen. You can in the stanbol working dir and
> search for the dbpedia.solrindex.ref file (find . -name
> "dbpedia.solrindex.ref"). This file is a Java properties file and the
> value of the "Directory" parameter will tell you what file is actually
> used.
>
> However please have also a look at the logs. Maybe this behavior is
> caused by some Exception during the initialization.
>
> best
> Rupert
>
> > Regards
> > Andrea
>
>
>
> --
> | Rupert Westenthaler             rupert.westentha...@gmail.com
> | Bodenlehenstraße 11                             ++43-699-11108907
> | A-5500 Bischofshofen
>

Reply via email to