issue with encrypt and decrypt

[Lily_Liu_2004]

DBAs,

We faced an issue with the encrypt and deencrypt with the obfuscate package.
The situation is like this:
prod database is 9.2.0.1, the dev is 9.2.0.6
both of them are al32utf8.
the nls_lang in both servers are set to al32utf8.
we took a export from the prod and import into the dev.
there are some columns like ssn and credit card number in one table
cannot be deencrypted correctly in the dev. but they function very well in prod. we got the errors:
ORA-28232: invalid input length for obfuscation toolkit
BTW,most of rows can be deencrpted correctly.
Only a small portion of rows cannot.
for example;
in dev;
SQL> select 'ODDW',t.act_cd, dump(t.act_cd) from exam_person t where t.exam_person_id = 17973529;

o[~O²j4 L‰Ëi
Typ=1 Len=15: 10,11,111,91,126,79,30,178,106,52,0,76,137,203,105

in prod;
o[~O²j4 L‰Ëi

Typ=1 Len=16: 10,11,111,91,126,79,30,178,106,52,0,76,137,203,105,240

it looks like in the dev, the last 240 got cut off.

We checked the export and import, The logs show they are done
in al32utf8 character set.

Any idea about this?

Thanks,

[Lily_Liu_2004]

FYT;
the dmp file's head:

$ cat erst_data_Jan23a2006 | od -x | head
0000000 3000 2e00 0000 0000 0000 0000 0000 0000
0000020 0007 2e2e 0000 0000 0000 0000 0000 0000
0000040 3002 6578 706f 7274 5f73 746f 7265 5f64
0000060

how to interpret from this to know the exact character set fro this dmp file?

[Lily_Liu_2004]

Anyone can help with this? Thanks.

[tamilselvan]

Isn't it a good idea to have same Oracle release of dev and prod particularly when you try to implement some complex algorithm?

Downgrade your dev env to 9.2.0.1
OR
Upgrade your prod to 9.2.0.6

Tamil