|
4.4 Changes
< Previous Next >
Re: Logout sequence corner cases
Byron Coetzee / Consultant <>
22 May 2006 7:59AM ET Hi,
your code logic must check for missing messages always, and then retry for logout again.
Cheers
Byron
> > > - Sorry for crossposting, but my question didn't attract much
> > > attention in "4.0 Session" -
> > >
> > > After having read the specification, I'm still having some trouble
> > > properly understanding the logout sequence. Maybe someone on the
> > > list can help me out.
> > >
> > > Consider the following situation:
> > >
> > > I am the logout initiator and I send my Logout message. However,
> > > when the counterparty responds with his logout message, I discover
> > > that it has a seqNo too high.
> > >
> > > What can I do now? Sending a ResendRequest will consume a new
> > > outbound sequence number, so my earlier Logout message will no
> > > longer be the last message in the connection.
> > >
> > > On page 32, the FIX.4.4 specification says: "Note counterparty may
> > > send a ResendRequest message if Logout message response has
> > > MsgSeqNum too high and then re-initiate the Logout process."
> > >
> > > What does "re-initiate the Logout process" mean in this context?
> > > Does it mean that my ResendRequest invalidates my earlier Logout
> > > message and that I will have to send a new Logout message by the
> > > time the acceptor has sent the necessary GapFills?
> > >
> > > kind regards, Erik van Zijst
> >
> >
> > Send a TestRequest message before Send Logout Message, So you can find
> > gap from the TestRequest's heartbeat.
>
> But then what if the acceptor sends another message after it responded
> to the TestRequest? This post-TestRequest message could still get lost.
Re: Logout sequence corner cases
Byron Coetzee / Consultant 22 May 2006 7:59AM ET
|
@FIXProtocol
