Handle reading of the input (object) from the kettle repository by getting the job-xml from the repository step attribute string and then re-hydrate the
job entry (object) with our already existing read method.
Handle reading of the input (object) from the kettle repository by getting the step-xml from the repository step attribute string and then re-hydrate the
step (object) with our already existing read method.
The RepositoryObjectInterface represents all objects that can come out of a repository, including
directories (RepositoryDirectoryInterface), elements such as TransMeta and JobMeta
(RepositoryElementMetaInterface), and metadata about elements (RepositoryElementMetaInterface).
Slaves don't need ALL the partitions, they just need a few.
So we should only retain those partitions that are of interest to the slave server.
Divide the number of partitions (6) through the number of slaves (2)
That gives you 0, 1, 2, 3, 4, 5
Slave 0 : 0, 2, 4
Slave 1 : 1, 3, 5
--> slaveNumber == partitionNr % slaveCount
rounds double f to any number of places after decimal point
Does arithmetic using BigDecimal class to avoid integer overflow while rounding
TODO: make the rounding itself optional in the Props for performance reasons.
This test reads a CSV file in parallel on all 3 slaves, each with 4 partitions.
This is a variation on the test right above, with 2 steps in sequence in clustering & partitioning.
It then passes the data over to a dummy step on the slaves.