Re: more than 1 submit_sm being sent on same data packet

On Jan 6, 2008 12:46 PM, Andreas Fink <[email protected]> wrote:
>
>
> On 04.01.2008, at 22:10, Juan Nin wrote:
> what I mean is 2 submit_sm PDU's on 1 TCP packet...
>
> seeing the traffic capture on Wireshark (ethereal) I see this:
> http://img221.imageshack.us/my.php?image=picture2cw2.png
>
> there you see sequence_number 23 and sequence_number 24 (both
> submit_sm) were sent on the same TCP packet
>
> when this happens I receive the Throttling error
>
> I would say this is a very bogous interpretation of TCP. TCP can split and
> reassemble data. If an application sends 100 SMPP packets in one, TCP could
> split it even in the middle of a SMPP packet. This is out of the control of
> any application and operating system dependent. If the other end can not
> deal with this its a very bogous system.
>
> However more likely, as Alex pointed out, you have the problem of windowing
> that as long as the previous packet is not being acke'd, the next wont be
> accepted. So the error is that you send faster than the other side wants it
> accepted and not the fact that both packets are in the same TCP packet.
>
> The throttling error is not wrong by itself but it can slow you down too
> much as well I guess. It should not make any sms lost or duplicated in any
> case.


yeah, i know, but it happens quite a lot, so it slows down too much..  :(

about changing max-pending-submits slows down too much too...
setting "max-pending-submits = 1" for example slows me to 1 sms/sec  :(

i'm using one of the patches for controlling throughput on SMPP, and
from my side it appears to work ok, i've enabled the miliseconds
logging and i see it ok...

there are still not immediate plans on fixing that directly in kannel, right?