Re: FW: Datagram stuck and then sent twice

Forwarded on behalf of <jb40-BGeptl67XyCzQB+pC5nmwQ@xxxxxxxxxxxxxxxx>

--------------------------


Hello Andy,

I have run a UDP Java Prog on my TINIs (its about 50 pieces now) What I 
have found is that Windows often receive a UDP Packet with 0 Byte length. 
This semms to be a well known problem.
So my windows prog checks first the length of the packet and drops it if it 
0 Byte.

Second remark. If my network is heavy loaded it starts at loads about 30%; 
my TINI try to send the Packets (TCP in that case); the TINI looses Bytes 
at the serial interface. If anybody has a suggestion for that please give 
me a mail.

Thats my experience with TINI 390 :-)

Klaus


-------- Original Message --------
Subject: [TINI] FW: Datagram stuck and then sent twice (25-Jun-2003 15:49)
From:    Andy_Turcotte-WJpxu4zzjar3fQ9qLvQP4Q@xxxxxxxxxxxxxxxx
To:      klaus.amling-dP05va+lwlQjWyDKOx2mBQC/G2K4zDHf@xxxxxxxxxxxxxxxx

>
> Could this problem be due to a missed interrupt?
>
> -----Original Message-----
> From: tini-admin-6tN4nzCoH/hBDgjK7y7TUQ@xxxxxxxxxxxxxxxx 
[mailto:tini-admin@xxxxxxxxxxx]On Behalf Of
> Andy Turcotte
> Sent: Wednesday, June 18, 2003 4:11 PM
> To: TINI-5ZoS9duIjYpBDgjK7y7TUQ@xxxxxxxxxxxxxxxx
> Subject: [TINI] Datagram stuck and then sent twice
>
>
>
> My java app receives and sends datagram packets to a Windows 2000 server.
> The server can randomly report (one in ~2000 sends) that it has received
> the same packet twice. (The server is checking a uniquely stamped 16-bit
> cyclic sequence number within each packet.)
> The server does not report this problem against similar code running on
> Windows.
>
> I've confirmed that the app is thread-safe and isn't throwing any
> exceptions.
> I've added debug checking to catch my app in the act but the checks
> don't report a problem.
> When the the problem occurs, there is a noticeable pause in network
> communications. The server reports the problem at the end of this pause.
> I've placed debug statements in my app before and after the send().
> They show the send occurring (and completing) at the beginning of the pause.
>
> >From this it appears that the packet is not leaving the TINI until
> the end of the pause, at which time two copies of it are being transmitted.
>
> I'm running version 1.11 of the TINI OS on a TBM390.
>
> Can the datagram be getting stuck momentarily within the TINI OS, and then
> be unstuck (twice) later by incomming traffic?
> _______________________________________________
> TINI mailing list
> TINI-6tN4nzCoH/hBDgjK7y7TUQ@xxxxxxxxxxxxxxxx
> To UNSUBSCRIBE, edit your profile, or see list archives:
> http://lists.dalsemi.com/mailman/listinfo/tini
> _______________________________________________
> TINI mailing list
> TINI-6tN4nzCoH/hBDgjK7y7TUQ@xxxxxxxxxxxxxxxx
> To UNSUBSCRIBE, edit your profile, or see list archives:
> http://lists.dalsemi.com/mailman/listinfo/tini

_______________________________________________
TINI mailing list
TINI-6tN4nzCoH/hBDgjK7y7TUQ@xxxxxxxxxxxxxxxx
To UNSUBSCRIBE, edit your profile, or see list archives:
http://lists.dalsemi.com/mailman/listinfo/tini