Hi. If I run this sqoop import as an import only (no job) and remove the incremental, I can use target-dir with hdfs.
In fact I notice that if I don't use any incremental, I can use tartget-dir with hdfs... (job or normal import) On Jul 2, 2013 11:47 PM, "Jarek Jarcec Cecho" <[email protected]> wrote: > Hi sir, > Sqoop requires Hadoop configuration files available on the machine where > you run Sqoop. I'm wondering if the config files from machine "C" (hdfs > gateway I suppose) are also available on machine "A" where Sqoop is running. > > Jarcec > > On Tue, Jul 02, 2013 at 07:02:41PM +0900, corbacho anthony wrote: > > Hi! > > > > I am trying to create a sqoop job with incremental option. > > I want to save it into my hdfs, so I use the option --target-dir, > > but sqoop throw me an error: tool.ImportTool: Imported Failed: Wrong FS: > > hdfs://my.hdfs.com:54310/job_import_incrt, expected: file:/// > > > > My sqoop job: > > sqoop job --verbose --create job_import_0 -- import --connect > jdbc:mysql:// > > db.mysql.com:3306/DB --table TABLE_TEST --target-dir hdfs:// > > my.hdfs.com:54310/db_import --username xxx --password xxx --incremental > > append --check-column id --last-value 1 > > > > I run sqoop on a machine A, I have sqoop-metastore on a machine B and my > > hdfs on a machine C. > > > > What should I do to "force scoop" to save it into my hdfs and not on my > > local machine? > > > > PS: If i change --target-dir with a local directory, its work like a > charm. > > > > Thank you > > Anthony >
