ORA-24365 on 9i
Actually, this is the question:
$ sqlplus system/sysdev@dev
SQL*Plus: Release 188.8.131.52.0 - Production on Mon Feb 10 22:10:56 2003
(c) Copyright 2000 Oracle Corporation. All rights reserved.
Oracle8i Release 184.108.40.206.0 - Production
JServer Release 220.127.116.11.0 - Production
SQL> select name from v$database;
SQL> create public database link ENKL.WORLD connect to SYSTEM identified by password using 'ENKLAV';
Database link created.
SQL> select name from v$database@enkl;
select name from v$database@enkl
ERROR at line 1:
ORA-02068: following severe error from ENKL
ORA-24365: error in character conversion
@enkl is a db link to 18.104.22.168 database.
The best part is: when I tryed to do the same from 9i it silently killed listener on 8i database. Does anybody have a clue what is this all about? Brief search on Metalink gave no results...
NLS_LANG=AMERICAN_AMERICA.UTF8 on all instances.
8i was created as UTF8/UTF8, 9.2 was created as AL32UTF8/UTF8
Re: ORA-24365 on 9i
The problem is the AL32UTF8 CS of the 9i DB.
The main workaround is to create the Oracle9i database with the CS of UTF8 which is recognized by Oracle 8i clients.
About the LISTENER: is it meant for both 8i and 9i databases?
About listener - listener was OK on both systems.
Then I executed query like :
insert into remote_8i_schema.table@link
insert * from local_9i_schema.table;
as a result i got a message regarding not null constraint violation (as i discovered later it was due to wrong character set conversion - all varchar2 data were transformed into nulls) and as an additional effect listener on 8i dissapeared.
No core file or trace or even record into any of log files were generated - listener just dissapeared like it was killed if kill -9.
I tried to check docs yesterday regarding this AL32UTF8 - oracle recommends it as a default one - what is the actual differene between it an old 8i style UTF8 character set? What are the benefits of this new AL32UTF8 and why the hell it`s not compatible with old UTF8?
I got a feeling that documentation says that this new crap was created to be compatible with everything one may have.
Oracle 8i is simply not aware of that new AL32UTF8. Someone told me that Oracle might give you a one time patch for that, you better open an iTar with Metalink.
UTF8 is not 100% unicode compatible. It stores wrongly the surrogate pairs for example.
UT8 will not be supported in 10i. That is Oracle suggests that all new databases be AL32UTF8.
I did make one new DB AL32UTF8 :-)