Closed tobegit3hub closed 5 years ago
On the README page we've mentioned:
libdisni.so must be in java.library.path on every Spark Master and Worker (usually in /usr/lib)
So if libdisni is in default jaava.library.path
on each node (e.g. in /usr/lib
) then no need to configurre additionally.
We have found some issue about the wiki page in https://github.com/Mellanox/SparkRDMA/wiki/Running-HiBench-with-SparkRDMA which may be fixed easily.
In section "Experiment #1: TeraSort", we need to add not only
spark-rdma-3.1-for-spark-SPARK_VERSION-jar-with-dependencies.jar
but alsolibdisni.so
.It should be changed to this.