RE: [Quickfix-users] Quick FIX reconnection problem?

[Brendan B. B. <br...@ka...>]

In my debugging, I see that the QuickFIX client is sending the cached
message but on the QuickFIX server (executor), in:

bool Session::nextQueued( int num )
{
...
    if ( msgType == MsgType_Logon )
    {
        m_state.incrNextTargetMsgSeqNum();
    }
}

Thus when the client resends the cached NOS, the server tosses it
because the MsgSeqNum is lower than expected.

If the above was commented out, then if the client were to issue a
Resend Request to the server following the logon, on the server in
Session::nextResendRequest() the target MsgSeqNum is reset as well,
also leading to the NewOrderSingle cached prior to the logon to be
discarded because the MsgSeqNum is lower than expected.

Does anyone know if this is expected behaviour?

Regards,
Brendan


-----Original Message-----
From: qui...@li...
[mailto:qui...@li...]On Behalf Of Pedro Sousa
Sent: Monday, August 25, 2003 11:03 AM
To: qui...@li...
Subject: [Quickfix-users] Quick FIX reconnection problem?


Hi,

I'm having the following problem using QuickFix:

We have an application using QuickFIX as an Acceptor. Another FIX
engine is connected to ours to send Stock Orders. Sometimes, the
connection between the FIX engines is lost. When the connection is
re-established, both FIX engines try to reconnect and send the
messages shown below. The problem is that message 619 (a
NewOrderSingle message) is lost! It is logged, but never reaches the
"Application.fromApp" method. It seems the Order is sent during the
reconnection process, but that shouldn't be a problem, right? And
what does the message "Value is incorrect (out of range) for this
tag" mean, in message 622? I've check incoming message 617 but I
can't see nothing wrong with it. Is there something wrong with the
reconnection messages sequence? I Could you please help me? This
happened 3 times in 3 consecutive days!

Thank you.

Pedro Sousa