Using neo4j 3.4.3 we're seeing some transaction timeouts using SDN for some bulk saving and we're curious if modifying the timeouts would help. Currently we have this configuration deployed everywhere:
What would be some sane defaults? Could the acquisition timeout be too aggressive?
Understanding transaction and lock timeouts
There is no general answer to this since it depends on some many aspects of the given use case & setup.
If you're doing large batched write transactions concurrently you'll more likely to run into lock contention than doing small transactions. But small txwill have a higher overhead and therefore be slower.