Re: Losing UDP packets with MFC Sockets

Yes that exactly what I am saying. If you are at 100% cpu usage, while two
packets arrive, then the next time you get a OnReceive notification and do a
Receive you will pick up both packets in one buffer. I had a similar
problem with TCP sockets, had to build a parser that would identify parse
multiple packets out of the buffer, and also build a packet out of multiple
receives.

AliR.

"Vicent Soler" <VicentSoler@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:EBCAC313-93F0-43B4-B0C5-86E5FD14DBAC@xxxxxxxxxxxxxxxx
> AliR,
>
> Are you saying that in the input buffer I read from ReceiveFrom could
exist
> more than one packet?? I've always thoght that after receiven one
> notificication I'm nly processing one packet....
>
> Vicent
>
> "AliR" wrote:
>
> > You still didn't answer the question. Is this senerio being handled
> > correctly?
> >
> > Say the sending process sends out 3 packets of 20 bytes each, while your
> > receiving process is busy doing other things.
> > Your receiving process does a Receive with a buffer of 100 bytes the
next
> > time it gets a OnReceive notificaiton. Which will receive all 60 bytes
worth
> > of data.
> > Is the data parsed out so that all three packets gets processed?
> >
> > AliR.
> >
> > "Vicent Soler" <VicentSoler@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
> > news:E82253A2-68DE-4A12-A438-5F099772D37F@xxxxxxxxxxxxxxxx
> > > Hi AliR,
> > >
> > > This is the process to retrieving data from socket's buffer:
> > >
> > > 1.- After OnReceive method, a call to ReceiveFrom method is done.
> > >
> > > 2.- A CByteArray object is creating with 'new' with data received from
> > buffer.
> > >
> > > 3.- The CByteArray is passed to an internal list of the socket class
which
> > > is later accesed by other classes.
> > >
> > > 4.- The process ends after sending a Windows' message to the
respective
> > > class to indicate the packet arrival.
> > >
> > > As you can see, we always try to avoid being excesive time in the
> > OnReceive
> > > method. We will try to introduce threads in the process in order to
solve
> > the
> > > problem.
> > >
> > > Thanks
> > >
> > > Vicent
> > >
> > > "AliR" wrote:
> > >
> > > > This might have to do with the way you are handeling your message
> > retrieval
> > > > from the receive buffer, are you handeling receiving multiple
message in
> > a
> > > > single read?
> > > >
> > > > AliR.
> > > >
> > > > "Vicent Soler" <VicentSoler@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in
message
> > > > news:3AAE5247-76AB-43A3-912A-61177FAA3A49@xxxxxxxxxxxxxxxx
> > > > > Responses below....
> > > > >
> > > > > "Michael K. O'Neill" wrote:
> > > > >
> > > > > > Please give us some more information about your UDP usage (e.g.,
> > packet
> > > > > > size, frequency etc). IIRC, CPU load is not the main culprit in
UDP
> > > > packet
> > > > > > loss; rather, it's network load.
> > > > >
> > > > >
> > > > > The UDP packet size is variable but usually it can be around
50-300
> > bytes.
> > > > > The frequency is also difficult to calculate because the UDP
Server
> > send
> > > > > messages using event triggers. As far as we could understand, the
> > main
> > > > > problem is when the UDP Client is using lot of CPT time. Then it
seems
> > > > that
> > > > > the input buffer is being overload and packets are being
discarded.
> > > > >
> > > > >
> > > > > > The usual way to reduce UDP packet loss is to favor smaller and
more
> > > > > > frequent packets over larger and infrequent packets. It is
often
> > said
> > > > that
> > > > > > packets around the size of the MTU are best. Also, reduce the
send
> > > > buffer
> > > > > > size and increase the receive buffer size.
> > > > > >
> > > > > > Mike
> > > > > >
> > > > > >
> > > > > > "Vicent Soler" <VicentSoler@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in
> > message
> > > > > > news:9758AB39-50C0-4750-8F48-2F5399232EEA@xxxxxxxxxxxxxxxx
> > > > > > > Hi all,
> > > > > > >
> > > > > > > We are developing a tool which uses UDP packets to receive
data
> > from a
> > > > UDP
> > > > > > > Server. The problem we have found is that some UDP packets are
> > being
> > > > lost
> > > > > > > when the PC's CPU is near 100% and we think that this problem
is
> > > > related
> > > > > > to
> > > > > > > the Window's input buffer.
> > > > > > >
> > > > > > > Any suggestion to solve this problem!! Is there any way to
change
> > the
> > > > > > input
> > > > > > > buffer size of the socket and store the received packet while
the
> > PC
> > > > is
> > > > > > > processing other data? Should we use threads to extract data
from
> > > > sockets?
> > > > > > >
> > > > > > > We are really worried about this problem because we can not
lose
> > so
> > > > much
> > > > > > > packets as we do.
> > > > > > >
> > > > > > > Using more than one port, cuould solve the problem?
> > > > > > >
> > > > > > > Thanks in advance,
> > > > > > >
> > > > > > > Vicent
> > > > > >
> > > > > >
> > > > > >
> > > >
> > > >
> > > >
> >
> >
> >


.

Relevant Pages

  • Re: Deaf CAsyncSocket on Windows Service.
    ... To read them out I need a buffer to ... the number of bytes waiting to be received. ... Perhaps the guy that said 'packets' should have said 'bytes of data' ... You get an OnReceive notification if there is input data to be read. ...
    (microsoft.public.vc.mfc)
  • Re: In-tree version of new FireWire drivers available
    ... Just to recap, the dual buffer receive mode, as described in section ... quadlet aligned amount of header data can be appended into one buffer ... *either* the header buffer or the payload buffer fills up. ... enough to hold headers for all the packets it takes to fill up the ...
    (Linux-Kernel)
  • Re: Deaf CAsyncSocket on Windows Service.
    ... To read them out I need a buffer to ... Short packets at the sender are coalesced into longer packets. ... THe receiver says "I have this much buffer space" and the sender is free to send ... You get an OnReceive notification if there is input data to be read. ...
    (microsoft.public.vc.mfc)
  • Re: Cisco 7204VXR Interface ignored packet
    ... Input queue drops (ignored packets) are usually due to buffers not being ... The buffer defaults are good 98% of the ... 256 max cache size, 256 in cache ... 22496866 hits in cache, 0 misses in cache ...
    (comp.dcom.sys.cisco)
  • Re: Persistent stall in the Cypress FX2 FIFO
    ... the FX2's firmware code, would be to install some sort of stub routines ... > how long they are (it appends even for 4 byte long packets). ... > a packet from the FIFO before sending the next one all works fine... ... smaller or larger than what a single buffer could hold. ...
    (comp.arch.embedded)