Logic for memory allocation

[mdrishti]

Hi,

I have made a local qlever sparql endpoint for my data, which combines two knowledge graphs (~23 GB of uncompressed ttl files; ~250,000,000 triples). Couple of simple join queries worked. But for some others, I keep getting errors like the following:

Tried to allocate 3.3 GB, but only 2.4 GB were available. Clear the cache or allow more memory for QLever during startup,

which I do (clear the cache and allow more mem - see config file below). But I am not quite sure 'what' memory is being referred to here - MEMORY_FOR_QUERIES, CACHE_MAX_SIZE or CACHE_MAX_SIZE_SINGLE_ENTRY ? Or is it something else?

I am pretty new to sparql and knowledge graphs. But I think I have enough memory to handle sparql queries on my server. Can't share the data as of now. Maybe, it is of not much help in the absence of data, but here is the config file :
https://github.com/digital-botanical-gardens-initiative/integrate_trydb_globi_enpkg/blob/master/qlever/Qlever.try_globi_v0-2_ttl

And here is the query I am trying to run:
https://github.com/digital-botanical-gardens-initiative/integrate_trydb_globi_enpkg/blob/master/sparqlQueries/q2.spql

I would appreciate if you can help me understand how to tune the parameters to avoid the memory allocation errors.
-DT