Re: [openss7] Hey

["Brian F. G. Bidulock" <bidulock@openss7.org>]

Chuck,

Chuck Winters wrote:                     Mon, 18 Jun 2001 18:06:40
> Brian,
>     Sorry, for the delay, but I was busy testing a protocol!  It does
> work.  The times seem to be good so far on a small sample size.  There
> are a couple of problems though.
> 
>     1)  sctp_fast_retrans_calc is not passing a debugging feature on
> line 202

See patch.

>     2)  recvmsg does not return anything in (struct
> msghdr).msg_name (it's null)

I believe it is normally null for TCP sockets.  On multi-homed SCTP
it might be an idea to return the actual sender, but it would still
be one of the addresses returned by getpeername().  It would cost
time to fill out this for ever message.  UDP fills this out.

It might be an idea to make this a socket option.

>     3)  Something locks up when there is a packet size of 4096.  It
> locks up after about 5 messages are sent
>           of that size

It would be good to get some ethereal dumps of this happening.  It
might have to do with SCTP attempting to fragment data chunks when
the chunks size is greater than 1/2 of the MTU.  (I presume that
you have an MTU of 5000 for etherenet.)

> 
>     I just started a run on 1,000,000 messages, so we'll see what
> happens.  The max. data size will be 2048 bytes, so it doesn't lock up.
> 
>     I still have to implement the rtt stuff into my code.  Maybe
> tomorrow I will start that so I can begin to tweak some settings.

I eagerly await results.  If you get any strange crashes, try to
grab an ethereal dump and any logs created.

-- 
Brian F. G. Bidulock    ¦ The reasonable man adapts himself to the ¦
bidulock@openss7.org    ¦ world; the unreasonable one persists in  ¦
http://www.openss7.org/ ¦ trying  to adapt the  world  to himself. ¦
                        ¦ Therefore  all  progress  depends on the ¦
                        ¦ unreasonable man. -- George Bernard Shaw ¦