Hi all,

Does anyone know whether recovery is possible if multiple writers
concurrently write to a single Hudi dataset? eg. if two duplicate Spark
applications are started up on accident that index to the same dataset.
Would the dataset be permanently corrupted/difficult to recover? As far as
I can tell this is undefined behavior, but I was wondering if Hudi can/will
rollback successfully if this happens. Currently using 0.6.0.

Thanks!
Max

Reply via email to