Closed migalkin closed 7 years ago
I expect that this happens with queries that result in a lot of intermediary results.
Could you try node --max_old_space_size=4096 -- bin/ldf-client …
?
Possible duplicate of #24.
One additional question:
All the LDF clients which received the rewritten queries
I assume you have used a federated query planner to split up the SPARQL query. Just as a side note, you know that the client also supports federation out of the box? (But maybe you were comparing different federation algorithms of course.)
@RubenVerborgh yes, we know that the client supports federation out of the box. We are testing a new federation technique and compare it to (including) LDF. How to specify this parameter for Docker container runtime?
CMD ["node", "bin/www"]
Great, curious to see those results :+1:
My colleague @rubensworks can help you with Docker.
@migalkin I'm not sure what you mean exactly with that Docker command parameter, could you elaborate? Perhaps a different issue should be created if this is not related to this client crash.
The readme contains two examples on how to evaluate queries with the Docker container.
@rubensworks The question is how we can start the LDF client with node --max_old_space_size=4096 -- bin/ldf-client …
in Docker.
You can simply override the Docker entrypoint when running, where you can append any of the regular options afterwards.
For example:
docker run -it --rm --entrypoint="node --max_old_space_size=4096 -- bin/ldf-client" ldf-client ...
@rubensworks Thank you for the suggestion, the following config worked for Docker:
CMD ["node", "bin/www", "max_old_space_size=4096"]
I'm running a C4 query from the set of Complex queries against Fedbench. The query is:
All the LDF clients which received the rewritten queries out of this crashed with the following errors. DBpedia:
Drugbank:
In other Complex queries two more clients crashed. Linkedmdb:
NYTimes:
I'd never expect it would crash with OOM error.