Dear Katerina,
just a question to clarify as you have probably repeated the version number of GDB by mistake: which is the VB’s version number? I guess it’s 9.0 or 9.1.0 right?
Anyway, has this happened only when you came back, or is this happening once in a while?
Kind Regards,
Armando
--
You received this message because you are subscribed to the Google Groups "vocbench-user" group.
To unsubscribe from this group and stop receiving emails from it, send an email to
vocbench-use...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/vocbench-user/5eecab18-9874-4761-abde-7b28693dd5e9n%40googlegroups.com.
Dear Armando,
I am sorry, you are right, in the installation I have vocbench is 9.1.0 and ontotext graph is 9.1.1. 😁
Anyway, I notice that I got this exception (ie SocketException:
Connection reset) after vacations.But it is not consistent. Today,
when I tried to access the project I got "Could not rollback RDF4J
transaction; nested exception is
org.eclipse.rdf4j.repository.RepositoryException:
org.apache.http.NoHttpResponseException: The target server failed
to respond". In both cases I noticed that some queries were
running in the graphdb, but in the latter case I couldn't access
the date (which is normally given the exception). In the
attachment files you could see the queries that run in graphdb. Do
you think that I need to increase the memory in graphdb to
increase its performance and maybe stop producing such errors? I
know that my dataset is quite big, thus I opted for an external
db, as it is suggested by the VB documentation.
Any help is welcome.
Best regards,
Katerina
"Could not rollback RDF4J transaction; nested exception is
org.eclipse.rdf4j.repository.RepositoryException:
org.apache.http.NoHttpResponseException: The target server failed
to respond"
Dear Katerina,
sorry for coming back so late. It might be, what is the size of the dataset (is it only that, or there are several hosted?) and what is the amount of memory given to GDB and ST?
Kind Regards,
Armando
Dear Armando,
As far as the configuration is concerned both GDB and Vocbench run as containers in a VM with 8GB memory and 4 CPUs. Both containers can allocate at maximum this memory. As far as the size of the dataset, which is the only hosted in the GDB, it contains 2,340,164 triples (which is just the basic subset of what our final lexicon will have). The n3 file has 44MB size.
I also checked
the memory and CPU consumption while I tried to open the dataset
from vocbench. And although I still got the initial error of "Could not rollback
RDF4J transaction; nested exception is
org.eclipse.rdf4j.repository.RepositoryException:
org.apache.http.NoHttpResponseException: The target server
failed to respond" and queries were running in GDB, the memory
consumption remained stable. Specifically GDB required
~2.5GB, while vocbench required ~800MB. The CPU consumption
spiked for the GDB to 2-3CPUs, while vocbench left to use the
single available CPU. Maybe more CPU is needed for the GDB ?
Any suggestions are welcome!
King regards,
Katerina