GDS gets stuck at Node2Vec :: train :: iteration 1 of 10 100%

Neo4j Graph Platform Graph Algorithms/Graph Data Science
1

Hey all. We're running node2vec on GDS with 16 cores, 100GB ram on a ~4M node KG.

However, the process always gets stuck on

2024-10-31 16:14:42.593+0000 INFO  [gds-4] Node2Vec :: train :: iteration 1 of 10 100%

it gets from 0 to 100% fairly quickly and then just sits there.

CPU is still fully utilizing all cores so it's doing something. But what? Can we get any more details from the DB? Any query we can run or logging we can tweak to know what happens after the 100% is reached?

2

Sorry for the very delayed response @pascalwhoop and for anyone who might be experiencing this

We have discovered for Node2Vec that our progress reporter tool was incorrectly reporting 100% for every iteration, even though there was still ongoing work for the iterations. We have addressed and improved progress reporting. The feature will be out with the next gds release (2.18.0).

Best regards and I apologize for the delay in resolving this issue,
Ioannis.