notebook for SFTP server connectivity without password.

[s3]

I am trying to develop some script using python to access an sftp server without password and all valid public/private keys in a notebook. However I am not getting any such example. All examples has a password in it. Can I get some help?

[Hubert-Dudek · Accepted solution]

@soumen sarangi​ , There is Spark SFTP library https://github.com/springml/spark-sftp

It sepends what is your data there. Sometimes better idea is to set some copy activity (like in Azure Data Factory) to have it in adls/blob mounted in databricks.

[skohade1]

I am currently using this library in my databricks where i am facing challenge to load the data from sftp

local_path = f'/Workspace/Users/<user.email>/Scard_etl/SFTP_LOCAL/JDL/'

sftp_df = spark.read.format("com.springml.spark.sftp")\

        .option("host", jdl_sftp.host)\

        .option("username", jdl_sftp.user)\

        .option("password", jdl_sftp.password)\

        .option("fileType", "txt")\

        .load("/dsdata/jdl/files/jdl_settlements.txt")

sftp_df.write.mode('overwrite').text(local_path)      
I am getting error below
Py4JJavaError: An error occurred while calling o504.load. : java.lang.NoClassDefFoundError: scala/Product$class at com.springml.spark.sftp.DatasetRelation.<init>(DatasetRelation.scala:24)

As per my research do I need to upgrade the version of library if yes where can find the upgraded version of this as i need to attach it to my cluster in databricks

com.springml:spark-sftp_2.11:1.0.3

[Atanu]

https://stackoverflow.com/questions/58562744/how-to-upload-text-file-to-ftp-from-databricks-notebook this example looks good to me. or may be try using data libs. https://www.cdata.com/kb/tech/sftp-jdbc-azure-databricks.rst

[Atanu]

again - I do not have this tested to my lab. so still unofficial. 🙂