hi, everyone,
I find that dgraph live load can specify UID in the rdf files, but dgraph bulk load cannot specify UID in the rdf files.
Is there any reason about this difference?
Maybe it’s more convenient for dgraph bulk load to support assigning UID.
@dgrdev The difference is simple. LiveLoad was created to feed data in a running Cluster. Bulk load was created for populate data when starting a cluster from scratch.
@relunctance I believe that this PR can bring problems. There is a situation about UID allocation (lease). If Bulkload allows the recognition of previously used UIDs can generate lots of errors during the load. It would be necessary to reallocate them in a cluster created from scratch. And you never know how many UIDs were actually alocated in a previous use. To find out, you would have to count in the RDF itself. This would delay the load a little.
This limitation is being worked out with the creation of the Dgraph Back-up. But for populate data in a New Cluster Bulk Load is the best choice.