Re: MYSQL error with IBMDB2 engine -- WEB400

Hi Dale,

I'm one of the storage engine developers, and I think I understand what
you're seeing. There was a bug in early releases of the storage engine
such that certain Unicode character sets were not being created with the
preferred CCSID (utf8 was created as CCSID 1200 instead of 1208, for
example.) This worked fine, if somewhat slowly, but we chose to fix this
behavior in later MySQL releases. Unfortunately, one side effect was that
tables that had been created with the previous behavior looked incorrect
to later versions, generating the 2501 error that you're seeing. You
didn't say that you had upgraded MySQL, but based on the time frames
indicated in your post, I'm guessing you were caught in this situation.

To work around this, you will need to use the older version of MySQL
(5.1.36 or earlier) to get the data out of any affected IBMDB2I tables and
then place the data back into the IBMDB2I tables using a current release
of MySQL. One way to do this is to use the older version of MySQL to alter
the tables to use MyISAM. Then upgrade to a current release of MySQL and
alter the tables back to use IBMDB2I again. Or you could select the table
data into an outfile under the old version and with a new version recreate
the tables (i.e. reinstall SugarCRM) and select the data from the outfiles
back into the tables.

As far as support goes, any failures that are clearly IBM i related (e.g.
produce CPF or MCH messages in a job log) can be reported directly to IBM
Support under your normal support terms. Bugs such as this one should be
reported to MySQL through your respective support channels (if you have
purchased the Enterprise edition of MySQL) or through the bugs.mysql.com
website (for Community edition users.)

Tim Clark

dale janus wrote:

We have been using PHP and MYSQL (and APACHE) for about 2 years. We
have been pretty successful created web based aps to view I series
data using the DB2 calls Zend put into PHP. We are running V5R4.

We were very happy to see that IBM finally created a database engine
for DB2 for MYSQL. We installed it back in May.

Now we are expanding and trying to create web based front ends for
our data entry programs. We set up PHP to use MYSQL with the DB2
engine to edit and enter a batch file for our billing system. At
first, it worked fine. Now we are getting errors:

Got error 2501 'Error opening codeset conversion from UTF-8 to
UTF-16 (errno = 0)' from IBMDB2

We could not chase it down, so we switched the engine back to MYISAM
in MYSQL and naturally it works perfectly.
For this particular application it does not matter what engine is
used, but we are getting the same error with SUGARCRM .

We installed SugarCrm back in May with the default test data using
the IBMDB2 engine in MYSQL. Sure enough, it created a I SERIES
library called "sugarcrm" and a bunch of physical and logical files,
all lower case with " "around them. I played around with it a while
back in May and then moved on to more pressing problems. Now I have
come back to it, and it starts with the error 2501. We had to change
the user file in SugarCrm MYSQL from IBMDB2 back to MYISAM. Then we
could log on. This removed the file from the I series library

"sugarcrm".

This allow SugarCrm to run, but whenever I try to view the existing
test data provided by SUGARCRM, I get an error messasge: Error
retrieving record. This record may be deleted or you may not be
authorized to view it.
This happens no matter if I am logged on as myself, a test user, or
administrator.

Here I am, finally getting ready to put my data into SugarCrm and it's

broken.

Is anybody trying to use MYSQL with native files? Has anyone run
into this problem?

Is this something I can call IBM Software Support about?

Some of the MYSQL forums indicate these causes for the error can be:

This is due to lack of support for the collations on v5r4

or

When a user selects an unsupported character set for an IBMDB2I table

It appears MYSQL will eventually release a patch, but all it does is
change the error message from internal error to user error.

---Dale