This APAR applies to Unicode databases using language aware
collation where a column of string type (for example: char,
varchar, clob types) contain invalid UTF-8 characters.
This APAR does not apply if there is no invalid UTF-8 character
in any of the columns. See
http://www.unicode.org/versions/Unicode6.0.0/ch03.pdf (Table
3-7) for well formed UTF-8 Byte Sequences.
If there are invalid UTF-8 characters in a Unicode database
created using language aware collation, the following errors can
occur:
1) SET INTEGRITY may fail with the following error:
SQL0901N The SQL statement or command failed because of a
database system error. (Reason "Key data mismatch encountered
during key delete".)
SQLSTATE=58004
2) DELETE of a row that exists in the table may fail with the
following error:
SQL0100W No row was found for FETCH, UPDATE or DELETE; or the
result of a query is an empty table. SQLSTATE=02000
3). Incorrect data is returned from the sort statement or data
is not sorted correctly.
This APAR is applicable if :
1. The Unicode database using language aware collation was
created on a previous release, migrated to v10.1.0, and had a
column using a string type that contains invalid UTF-8
characters.
2. A new Unicode database is created using language aware
collation on v10.1. 0 and one of columns is using a string type
contains invalid UTF-8 characters.
The issue affects database collation. If there is no other
object that depends on that column, fix the invalid character to
get the correct sort order in the query processing. If the
database contains other objects, such as indexes or MQT, which
depend on the column containing invalid characters, then drop
the database object(s) that depend on the object with invalid
characters. Apply the APAR. Recreate or refresh the affected
object(s). Rebind the static packages.
In the following example, a hex value is being inserted that
denotes an invalid UTF-8 character, resulting in the server
failing to delete the row:
db2 "CREATE DB WSDB USING CODESET UTF-8 TERRITORY US COLLATE
USING SYSTEM_819_US"
db2 connect to wsdb
db2 "create table t1 '(c1 varchar(20))'"
db2 "insert into t1 values ('ROMANIA'), ('SAINT'),
(x'52C9554E494F4E')"
db2 "create index i1 on t1 (c1)"
db2 delete from t1 where "c1=x'52C9554E494F4E'"
SQL0100W No row was found for FETCH, UPDATE or DELETE; or the
result of a query is an empty table. SQLSTATE=0200
To fix the issue in the example: Drop the index, apply the APAR,
and recreate the index. |
