Re: [Quickfixj-users] One More Time

[Steve B. <st...@te...>]

Hi Peter,

 

Can you be more specific about the failed return from SystemTime.getDate()?
If the time source

is set correctly (which it is, by default) this is ultimately just a call to
System.currentTimeMillis().

Is there a stack trace?

 

I know you didn't like my suggestion to reproduce this behavior with Banzai
and the Executor,

but if you could reproduce it in that context I'd be able to investigate the
problem in more

depth by running it myself. If it really is an acceptor problem, then any
client should be able

to trigger it.

 

Steve

 

  _____  

From: qui...@li...
[mailto:qui...@li...] On Behalf Of Peter
Warder
Sent: Sunday, November 12, 2006 9:54 AM
To: qui...@li...
Subject: [Quickfixj-users] One More Time

 

I still can't get my QFJ acceptor to stay running! I have an associate in
Munich who is also trying to set up a QFJ acceptor and is getting exactly
the same problem as we are in London.

I have already forwarded a detailed composite of message and event logs to
the mail list. No response to that. 

However, I have now trawled through the source code and think I can see
where the problem is arising (but not why).

 

I've looked through Session.java and the behaviour we see is consistent with
when a reset() is called.

The reset() appears to be triggered by a false return from
checkSessionTime().

This, in turn, looks like a failed return from SystemTime.getDate.

 

Does this help anyone point us to the problem?

 

Also, when our acceptor fires up, it creates a .seqnum file which is zero
length and obviously empty. This never seems to be initialised although the
incoming logon has its sequence number in it. (Perhaps, this is because the
reset() forces a logout and disconnect - so we never get started.)

Should we set up a default start up .seqnum file with zeroes/ones in it? If
yes, where do we find the file format?

 

Please try to get us going on this one guys.

 

Peter Warder

London