Possible characterset issues w/ datapump
- From: "Don Seiler" <don@xxxxxxxxx>
- To: oracle-l <oracle-l@xxxxxxxxxxxxx>
- Date: Fri, 19 Jan 2007 15:58:57 -0600
We're having an issue where a "ú" character in a varchar2 is being
corrupted after being funnelled through datapump. I'm reluctant to
blame anything right now. Here is some background on the two
machines, SOURCE and DEST. Both machines are on RHEL 3.
SOURCE:
Oracle 10.2.0.2
OS $LANG = en_US
OS $NLS_LANG = AMERICAN
NLS_CHARACTERSET = US7ASCII
DEST:
Oracle 10.2.0.1
OS $LANG = en_US.UTF-8
OS $NLS_LANG = AMERICAN
NLS_CHARACTERSET = WE8ISO8859P1
From Oracle SQL Developer on my laptop, when I query the field with
the funny character on SOURCE, I just see a square. When I query the
field on DEST, I see it properly. Here is where things get
interesting.
We use expdp to dump a file from SOURCE, then FTP that file over the
DES, then impdp that file into DEST. The developers say that the "ú"
gets corrupted somewhere in that process, and that the hex value of
the field is different than on SOURCE. When I said earlier that it
looked fine on DEST, it was after they ran an UPDATE zap to correct
the data. I'm waiting for another trial to see the bad data for
myself, so bear with me.
Given that the DEST characterset is actually a superset of SOURCE, I
wouldn't expect any corruption or dataloss. Is there anything else I
can look at or specify to track down when and where any corruption is
taking place?
--
//www.freelists.org/webpage/oracle-l
Other related posts:
