I'm having a lot of trouble remembering why we won't let anyone close
a connection with an active transaction. At the moment it seems to
me that if the tx has not been prepared we should roll it back and
close the connection, otherwise just close it and the tx will be in
limbo.

Can anyone remember why we made the decision we did? I think it made
sense at the time :-)

thanks
david jencks

On Mar 14, 2007, at 5:00 PM, Roman Rokytskyy wrote:

> > I am currently forcing a fatal error in FBManagedConnection.start
> () when it cannot start a transaction to try and force jboss to
> discard the connection. I will run this tonight and see if it
> helps. I guess that Jboss's transaction manager should discard a
> connection it cannot create a transaction on, But I am hoping that
> the XA_UNKNOWN(0) error code may be confusing it.
>
> I think this is ok for now, but I will check whether better
> solution is
> possible.
>
> > I also noticed that are a number of places in with managed
> connections where we throw isc_network errors which are now fatal
> but because the function is called via the Managed connection
> rather then GDSHelper we do not notify anyone that the connection
> is potentially dead. I am wondering if this is likely to be a
> problem. Specifically in areas where we try and recover
> transaction. In this area we are using GDS_java_impl directly.
>
> This is bad. I will review such places, and use some different error
> code in this case.
>
> Thanks!
> Roman
>
>



[Non-text portions of this message have been removed]