Hi Lorenz,

When I watch the memory consumption in top java is running steady at 39g
for about 15 minutes (with -Xmx24g -Xms24g set).
With max. memory set to 43g docker exits with code 137 (out of memory),
shortly after startup (with inference=false):

This is not the behaviour when *--inference* is set (fuseki tx abort).
We have to use inference to access the alternate geometry predicates in our
dataset(s).
They don't contain the right geosparql predicate.

Andy mentioned earlier that the following error messages should/could be in
the logging, but I didn't see them:



*Should be a log line: "Inferencing Error: ..." as well or there is ajava
Throwable/Error happening e.g. OutOfMemoryError.   *


Op vr 11 feb. 2022 om 10:17 schreef Lorenz Buehmann <
buehm...@informatik.uni-leipzig.de>:

> I'm pretty sure the inference will be computed in-memory, so maybe it's
> a memory issue?
>
> On the other hand, do you really need the inferences computed for the
> GeoSPARQL schema? If not, it's possible to disable it in the assembler
> file.
>
> On 11.02.22 10:08, Erik Bijsterbosch wrote:
> > Hi Adrian,
> >
> > Thanks.
> > I noticed this repo yesterday.
> > I will check it out later.
> >
> > For now I really want to know what's happening in my implementation,
> > which works fine on my local machine with a small dataset.
> > Some more tests show that fuseki only aborts while *inferencing *my 256
> > million triple dataset.
> >
> > Maybe someone can provide me with a way to debug this...
> >
> > Regards,
> > Erik
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > Op vr 11 feb. 2022 om 06:29 schreef Adrian Gschwend <ml-...@netlabs.org
> >:
> >
> >> On 10.02.22 22:18, Erik Bijsterbosch wrote:
> >>
> >> Hi Erik,
> >>
> >>> I pursued my attempt to set up a dockerised fuseki-server and
> >>> fuseki-geosparql combi application.
> >>> I created one image for boh services which I can start with
> >> docker-compose
> >>> arguments.
> >> You might want to try the work my colleague Ludovic did:
> >>
> >> https://github.com/zazuko/fuseki-geosparql
> >>
> >> regards
> >>
> >> Adrian
> >>
>

Reply via email to