RE: reconnect failing in 3.51.12

A final note on this issue - I noticed I didn't have the problem at all on my SuSE 10.2 machine.

 

In order to truely keep my ODBC client app from crashing during reconnect, I had to upgrade to the following revisions of rpm packages on my SuSE 10.1 box:

 

 

   unixODBC-2.2.12-13.i586.rpm (from 2.2.11-19)
   mysql-5.0.26-12.i586.rpm (from 5.0.18-20.8)
   mysql-client-5.0.26-12.i586.rpm (from 5.0.18-16)
   mysql-shared-5.0.26-12.i586.rpm (from 5.0.18-16)
   MyODBC-unixODBC-3.51.12-33 (from 3.51.12-11)

 

 

The upgrade of unixODBC did nothing. Upgrading to mysql 5.0.26 and MyODBC 3.51.12-33 is what fixed the problem for me. I'm now able to completely reconnect with no trouble.

 

I checked the ChangeLog on MyODBC 3.51.12-33 - apparently the only difference between 3.51.12-11 and 3.51.12-33 is that make -j was used to build (multi-threaded make). Should have been a harmless build process change. I believe ultimately it made no difference.

 

I'm guessing the problem was fixed between mysql 5.0.18 and 5.0.26.

 

Question: Why couldn't someone from MySQL have told me this? It's not like there's a lot of traffic on this list. Don't they monitor?

 

John

 

-----
John Calcote (jcalcote@xxxxxxxxxx)
Sr. Software Engineeer
Novell, Inc.


>>> On Mon, Feb 12, 2007 at  4:08 PM, in message <45D09106.37FF.0081.0@xxxxxxxxxx>, "John Calcote" <jcalcote@xxxxxxxxxx> wrote:

Just wanted everyone to know the resolution to this problem. There's an option that can be specified in the (unixODBC) odbc.ini file to enable an option called "auto-reconnect". This option supposedly allows transparent reconnection with a few caveats.

 

   (/etc/unixODBC/odbc.ini)

   ...

   options = 4194304

   ...

 

The value 4194304 is decimal for the hex value 0x400000, which makes a little more sense - it's a bit flag that indicates an option to the myodbc driver.

 

Using this option, my ODBC client app no longer crashes (yet). To me this is quirky behavior anyway because this option should not be related to the way I was reconnecting anyway. In reality, it's a work-around - allow the underlying driver to reconnect rather than doing it in the app.

 

I'll let you all know if this status changes.

 

Thanks for the suggestions,

John

 

 

-----
John Calcote (jcalcote@xxxxxxxxxx)
Sr. Software Engineeer
Novell, Inc.

John Calcote.vcf
Description: Text document

-- 
MySQL ODBC Mailing List
For list archives: http://lists.mysql.com/myodbc
To unsubscribe:    http://lists.mysql.com/myodbc?unsub=gcdmo-myodbc@xxxxxxxxxxx