[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Raw Post/Connect Issue
Thanks to all who chimed in on this. The vendor and I went around, and
they actually shipped a laptop and test device to me so they could
remote in an check out what was happening. The developer I was working
with was making changes to their C (not ++, not #) as we were speaking.
It turns out that they weren't receiving the entire transmission, and
were moving on and causing an error. They basically put retries in to
read the entire message I was sending.� I'm using a sockets program (thanks Scott) instead of an HTTP POST
(thanks Scott) and it's working pretty well.� Again, thanks for all that had ideas and shared them!
On Mon, Mar 16, 2015 at 9:03 PM, Charles Wilt
<[1]charles.wilt@xxxxxxxxx> wrote:
� �Yep going back to the raw comm traces, comparing the trace from
IBM i
� �<-->Device to PC <-->Device will give you the answer.
� �You might have to dig into the raw bytes sent instead of just
looking
� �at the pretty wireshark summary...
� �You also might see if there's anyway to trace from the device
side.
� �I once got involved in trying to figure out why we were having
comm
� �errors on Win Mobile 5.0 cellular handhelds talking to a
backend app
� �running on a windows server.� And just for fun, there was a
"relay"
� �server in the DMZ that the connections got relayed through.
� �Traces at the device, the relay server and the backend server
told the
� �story...
� �Turned out that while the software used a protocol identified
as HTTP,
� �it wasn't really.� When the packets would hit certain parts
of the
� �Verizon network, they would get broke up/reformated/optimized.� � �Perfectly legitimate for HTTP but the app could not deal with
it.� Had
� �to move to a different port so version would leave them alone.
� �Good luck!
� �Charles
� �On Mon, Mar 16, 2015 at 6:22 PM, Mike Krebs
� �<[1][2]mkrebs@xxxxxxxxxxxxxxxxxx> wrote:
� � �I am sure you understand that if the device is in error at
connect,
� � �something is really wrong. But could the TCP connect code be
bad on
� � �your IBMi? I doubt it.
� � �Back to comparing the wireshark SYN/ACK-SYN/ACK bytes and
see if you
� � �can determine what the IBMi SYN/ACK-SYN/ACK is sending or
not
� � �sending that the device doesn't like.
� � �I can't imagine the device people wrote their own sockets
but if so,
� � �they missed something. If not, could it possibly be wrong?
� � �Unlikely....but you know one or the other is wrong.
� � �Hmmm? Fun debug - and I'm sure pretty frustrating at this
point.
� � �Good luck.
� � �-----Original Message-----
� � �From: [2][3]ftpapi-bounces@xxxxxxxxxxxxxxxxxxxxxx
� � �[mailto:[3][4]ftpapi-bounces@xxxxxxxxxxxxxxxxxxxxxx] On
Behalf Of
� � �Michael Ryan
� � �Sent: Monday, March 16, 2015 2:59 PM
� � �To: HTTPAPI and FTPAPI Projects
� � �Subject: Raw Post/Connect Issue
� � �I had the post about using http_raw to post to a device. The
vendor
� � �has been pretty accommodating, but doesn't see the problem.
They
� � �asked me to write a socket program as a proof of concept
type thing.
� � �I took one of my existing programs (in solid production for
years),
� � �stripped it down, and gave it a try.
� � �When I do a tconnect (I've also tried a connect), the device
flashes
� � �communication error and puts a zero on the screen. I think
this is
� � �probably the same issue when I was using HTTPAPI. I receive
a return
� � �code of zero for the tconnect which I think is good. The
vendor is
� � �baffled since their Intel customers (everyone except me)
doesn't
� � �have a problem.
� � �Any stabs in the dark? Thanks...
� � �------------------------------------------------------------------
--
� � �---
� � �This is the FTPAPI mailing list.� To unsubscribe, please
go to:
� � �[4][5]http://www.scottklement.com/mailman/listinfo/ftpapi
� � �------------------------------------------------------------------
--
� � �---
References
� �1. mailto:[6]mkrebs@xxxxxxxxxxxxxxxxxx
� �2. mailto:[7]ftpapi-bounces@xxxxxxxxxxxxxxxxxxxxxx
� �3. mailto:[8]ftpapi-bounces@xxxxxxxxxxxxxxxxxxxxxx
� �4. [9]http://www.scottklement.com/mailman/listinfo/ftpapi
--------------------------------------------------------------------
---
This is the FTPAPI mailing list.� To unsubscribe, please go to:
[10]http://www.scottklement.com/mailman/listinfo/ftpapi
--------------------------------------------------------------------
---
References
1. mailto:charles.wilt@xxxxxxxxx
2. mailto:mkrebs@xxxxxxxxxxxxxxxxxx
3. mailto:ftpapi-bounces@xxxxxxxxxxxxxxxxxxxxxx
4. mailto:ftpapi-bounces@xxxxxxxxxxxxxxxxxxxxxx
5. http://www.scottklement.com/mailman/listinfo/ftpapi
6. mailto:mkrebs@xxxxxxxxxxxxxxxxxx
7. mailto:ftpapi-bounces@xxxxxxxxxxxxxxxxxxxxxx
8. mailto:ftpapi-bounces@xxxxxxxxxxxxxxxxxxxxxx
9. http://www.scottklement.com/mailman/listinfo/ftpapi
10. http://www.scottklement.com/mailman/listinfo/ftpapi
-----------------------------------------------------------------------
This is the FTPAPI mailing list. To unsubscribe, please go to:
http://www.scottklement.com/mailman/listinfo/ftpapi
-----------------------------------------------------------------------