0 votes
I have an old database (Oracle) that I changed to Vertica, and I have many flows that use the Oracle db tables and I want to replace only the the dataset without losing all the stream.

1 Answer

+1 vote


Most recipes, but not yet all, support replacing an input or an output.

The best solution, though not officially supported, is to use our API to change the settings of the existing datasets.

IMPORTANT NOTE: Generally speaking, just changing the type of a dataset will not magically work. However, in the specific case of SQL datasets, they are fairly easily substitutable: you just need to change the "type" and "connection" params.

The following code (which can run in a notebook) would do this (you have to go to Settings to add a new API key)

from dataikuapi.dssclient import DSSClient

host= "http://localhost:10000"

client = DSSClient(host, apiKey)
p = client.get_project(project)

for dataset in p.list_datasets():
	print "Checking dataset %s" % dataset["name"]
	d = p.get_dataset(dataset["name"])

	definition = d.get_definition()

	if definition["type"] == "Oracle":
		definition["type"] = "Vertica"
		definition["params"]["connection"] = "my_vertica_connection"
		print "Updating dataset %s" % dataset["name"]


1,082 questions
1,126 answers
10,892 users

©Dataiku 2012-2018 - Privacy Policy