Moved from GitHub dgraph/4829
Posted by larvinloy:
What you wanted to do
Given an existing cluster, we want to be able to insert relatively big data-sets (multiple files) that might contain nodes that already exist in the graph.
What you actually did and why it wasn’t great
To do this we decided to use
upsert blocks, but unfortunately it appears that I cannot pass an array of
upsert blocks to the live loader, the way you could with
Any external references to support your case
The only other option seems to be to adopt his approach here, but the number of connections we need to make between the nodes is huge and having to make each connection though an individual call to the mutate endpoint is not feasible.