MySQL Bugs: #32179: aix52 5.0.50 32-bit binary without large file support

Bug #32179	aix52 5.0.50 32-bit binary without large file support
Submitted:	8 Nov 2007 9:21	Modified:	17 Nov 2008 10:40
Reporter:	Domas Mituzas	Email Updates:
Status:	Closed	Impact on me:	None
Category:	MySQL Server: Compiling	Severity:	S2 (Serious)
Version:	5.0.50	OS:	IBM AIX (5.2 32-bit)
Assigned to:	Joerg Bruehe	CPU Architecture:	Any

Description:
AIX 5.2 32-bit binaries do not have large file support:

InnoDB: Error: file size must be < 4 GB with this MySQL binary
InnoDB: and operating system combination, in some OS's < 2 GB

How to repeat:
bin/mysqld_safe --innodb_data_file_path=ibdata:4300M

(or of course, create large myisam file :)

Suggested fix:
repackage

The suggestion to "repackage" is a bit incomplete,
any such change requires a new compilation.

The whole issue looks to be closely related to bug#31254
    "Max_data_length" truncated / reported wrong (compiler issue ?)
whose fix is pushed into the 5.0.52 sources.

I tried 5.0.51 (AIX 5.2, 32 bit) on our AIX 5.2 host 
(binaries were still available, sources match those of 5.0.50)
using the proposed command line
   bin/mysqld_safe --innodb_data_file_path=ibdata:4300M
and got the diagnostic message as given in the report.

I then tried 5.0.52 ("enterprise", AIX 5.2, 32 bit) on that host
using the same command, and did *not* get that line.

The server did not really start, 
because the user's file size limit is set to about 1 GB,
but it began creating the data file which it had not done for 5.0.51.

I set this bug to "Need feedback",
asking the reporter to re-check with 5.0.52
(builds are available internally, just not completely checked and published)
but I very strongly assume this issue is also fixed with bug#31254.

Correction to my previous entry:
Specifying a non-conflicting port explicitly, I *could* start the server,
just ran into the file size limit enforced by the OS user settings.

No feedback was provided for this bug for over a month, so it is
being suspended automatically. If you are able to provide the
information that was originally requested, please do so and change
the status of the bug back to "Open".

No feedback was provided for this bug for over a month, so it is
being suspended automatically. If you are able to provide the
information that was originally requested, please do so and change
the status of the bug back to "Open".

According to my tests (see above), 5.0.52 has solved this.

Checking the support case from which tis bug originated:
On December 17, 2007, the customer was asked to verify 5.0.52 solves the issue;
there is no information following the request, but the support case got closed automatically.

So I close this bug.