Hi,

We are having some issue while sftping the db2 udb exported file [del ascii].

db2 export - generates the file like

123
345
678

but after sftping the file to mainframe system, the layout looks like this
123456678.

It seems mainframe only support EBCDIC code. how do we take care of this? has anyone face same issue? any pointer and help will be appreciated.

source serve: aix
target: sfpt file put file to aix server and mainframe pulls the file from sftp server.
We are using DB2 UDB 8.2

Thanks
Suraj

EBCDIC is not the problem or you wouldn't see such numbers but rather completely garbled characters. I think that your "sftp" client or the "sftp" server looses end-of-line characters.

Is it an option to directly import the file from AIX into the DB2 z database?

Thanks a lot or quick response..

not, looks like mainframe frolks are copying file from sftp/unix server to mainframe using mvs/jcl program. I don't think there is an option to directly to mainframe system?

Does db2 udb export have any additional features to specify ebcdic [newer version of db2]?

Thanks
suraj

I don't think so.
I suppose sftp is taking the complete (ASCII) file as a single "record", i.e. with embedded newlines, and converting that to EBCDIC.
So on mainframe you will most likely receive something like "123 456 789", where the "blanks" are not EBCDIC blanks but "non-displayable characters" representing the ASCII LF character in EBCDIC.

I'm not familiar with sftp on mainframe, but maybe it has an option to generate multiple records instead of a single one with embedded line-feeds. I guess this option is the "AS TEXT" option, as opposed to "BINARY". (With FTP one would use the "TYPE ASCII" command, not the "TYPE BINARY".)

Also, make sure that the dataset into which sftp is writing has the correct record attributes, especially make sure the record type is "F" or "V", not "U".