My question was:
>
> Hello,
>
> we have an AS 2100 5/250 1 GB 3.2c with fddi as the server sending
> and an AS 600 5/333 256 MB 3.2g with fddi as the client receiving data
> both machines are connectet directly with fddi (both at fta0).
> we wrote two C-programs with socket open und list, connect and all that
> stuff..
> in the center there is an loop with (BUFFER=4312 & datastorage=huge MBs)
> gotbytes = recv(socket_descriptor , datastorage, BUFFER,0 );
>
> with little packages of data smaller than 4 MB there is no problem
> with the transfers...
> above 4-5 MB with raising probabilty the transfer will fail in the middle
> and gotbytes will show up -1 which means: EINTR = interrupted system call.
> no data could received.
> On our investigation we noticed, the bigger the receiving process is, the
> probability of failing is rising!
> Also we noticed a difference between 3.2C and 3.2G fddi_config!
> on 3.2C the command fddi_config -i fta0 -v 4.00 works ok
> on 3.2G the same will produce an error!!! ???
> My Questions:
> will we need a patch for the fddi?
> will we need to update fddi-firmware? (Rev. 2.46 )
> monitor net is showing up, that the rate of incoming packets is twice as
> high as outgoing on the receiving machine! Is that normal?
> we tried at first the BUFFER=4352 but always noticed: only 4312 bytes were
> transportet (gotbytes). Is there an explanation for this behavoiur?
> what is the best setting for ipmtu and BUFFER?
> shall we use setsockopt and manipulate on SO_RCVBUF ( 32K present 128K max) ?
> any help would be great! a summary will be published for sure!
> PPS: i once red in the news about FDDIMON, a tool for monitoring FDDI-Lines
> is that available for DU? and where to FTP?
>
the answer is:
a) yes: there is a patch for fddi_config (3.2C and 3.2G)
b) no: no firmware-update needed for the fddi-cards
c) yes: doubled rate seems to be normal behaviour
d) -
e) -
f) no infos about FDDIMON
and:
there is kernel-parameter you can change if you use TWO FDDI-Cards
to accelrate the FDDI-switching of the CPU...
OK...
Panther
>
Received on Wed Apr 16 1997 - 07:19:40 NZST