I were trying to create index for node's property :
create index process_pid if not exists for (n:Process) on n.pid
It take about 10 minutes in 100 million nodes. However, in the case of 640 million nodes, it take 10 hours, what made so difference? All setting (neo4j.conf...) stay the same in both 100 and 640 million nodes but only the number of node that needed to create index are different.
I had tried what here suggested, i.e., adding the following into neo4j.conf.
dbms.jvm.additional=-Dorg.neo4j.kernel.impl.index.schema.GenericNativeIndexPopulator.blockBasedPopulation=true
but nothing change.
I tried to change the neo4j.conf setting. In 100 million case, all my heap.initial.size, heap.max.size and pagecache.size are all default value, in 640 million case I setting as 31G,31G and 467G, respectively (command line suggestion), but also, nothing change (still very long time to create index).
Remark : CPU 44 Core 88 threads and Memory 512 GB in my computer. Neo4j version : enterprise 5.3. I use command line open neo4j via neo4j console in bin folder (neo4j server).
- Why make this and how to fix it? Thanks.
