You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@jena.apache.org by Davide <ka...@gmail.com> on 2018/06/30 08:32:00 UTC
Fuseki and bulk loading problem via HTTP (heap space)
Hi,
I have some problems with Fuseki server and the triples loading via HTTP.
In my code, I add triples in the local model. When it reaches 300'000
triples, I load the model in the database with HTTP using:
"RDFConnectionFactory.connect("url")"
to create a connection with the database. Then I load the model in Fuseki
with:
"conn.load(model)"
After that, I create a new model with "ModelFactory.createDefaultModel()",
and I start another time to load triples in the model. This approach works
iteratively.
Well.. for the first iteractions, the db loads data without problems. The
problem is when the algorithm has already load many models. Because the
loading becomes slower, until the server crash due to heap space problems.
I increase the heap space size until 8 gb but I haven't resolve the
problem. Then I would ask you how to resolve this problem, and if this
approach is correct for this kind of problems.
Regards
Davide
Re: Fuseki and bulk loading problem via HTTP (heap space)
Posted by Davide <ka...@gmail.com>.
I resolved the issue, It was a configuration problem. Thanks
2018-06-30 17:57 GMT+02:00 Andy Seaborne <an...@apache.org>:
> Davide,
>
> A couple of things aren't clear:
>
> * Is the server being used for other requests at the same time?
> * After how many interactions do problems occur?
> * What's the stacktrace for the heap space problem?
>
> Andy
>
>
>
> On 30/06/18 09:32, Davide wrote:
>
>> Hi,
>> I have some problems with Fuseki server and the triples loading via HTTP.
>> In my code, I add triples in the local model. When it reaches 300'000
>> triples, I load the model in the database with HTTP using:
>>
>> "RDFConnectionFactory.connect("url")"
>>
>> to create a connection with the database. Then I load the model in Fuseki
>> with:
>>
>> "conn.load(model)"
>>
>> After that, I create a new model with "ModelFactory.createDefaultMod
>> el()",
>> and I start another time to load triples in the model. This approach works
>> iteratively.
>>
>> Well.. for the first iteractions, the db loads data without problems. The
>> problem is when the algorithm has already load many models. Because the
>> loading becomes slower, until the server crash due to heap space problems.
>> I increase the heap space size until 8 gb but I haven't resolve the
>> problem. Then I would ask you how to resolve this problem, and if this
>> approach is correct for this kind of problems.
>>
>> Regards
>>
>> Davide
>>
>>
Re: Fuseki and bulk loading problem via HTTP (heap space)
Posted by Andy Seaborne <an...@apache.org>.
Davide,
A couple of things aren't clear:
* Is the server being used for other requests at the same time?
* After how many interactions do problems occur?
* What's the stacktrace for the heap space problem?
Andy
On 30/06/18 09:32, Davide wrote:
> Hi,
> I have some problems with Fuseki server and the triples loading via HTTP.
> In my code, I add triples in the local model. When it reaches 300'000
> triples, I load the model in the database with HTTP using:
>
> "RDFConnectionFactory.connect("url")"
>
> to create a connection with the database. Then I load the model in Fuseki
> with:
>
> "conn.load(model)"
>
> After that, I create a new model with "ModelFactory.createDefaultModel()",
> and I start another time to load triples in the model. This approach works
> iteratively.
>
> Well.. for the first iteractions, the db loads data without problems. The
> problem is when the algorithm has already load many models. Because the
> loading becomes slower, until the server crash due to heap space problems.
> I increase the heap space size until 8 gb but I haven't resolve the
> problem. Then I would ask you how to resolve this problem, and if this
> approach is correct for this kind of problems.
>
> Regards
>
> Davide
>