SCTP Discussions

Text archives Help


Re: [SCTP-Discussion] Questions with Socket API


Chronological Thread 
  • From: Michael Tuexen <Michael.Tuexen AT lurchi.franken.de>
  • To: discussion AT sctp.de
  • Subject: Re: [SCTP-Discussion] Questions with Socket API
  • Date: Wed Oct 10 21:51:02 2001
  • List-id: SCTP Discussions <discussion.sctp.de>

Dear Dilip,
see my comments below.
Best regards
Michael
On Wednesday, October 10, 2001, at 09:29 Uhr, Dilip wrote:

Hi

I have implemented a simple echo server and a echo
client. using SCTP socket API..
It works fine.

Examples like this will be included in the next release.
Nice to see that also other people use similar things
to test/understand things.
Had few questions..

1. If I connect to the server (has multiple interface)
DO I have to specify all the interfaces on that server
during ext_connect on client side.
IF NO then will the ext_connect() automatically check
for availabilty of service on ALL server interfaces?

The client does not has to specify all addresses. Only
one server address has to be specified. During the setup
of the association the client 'learns' the other addresses.
This is the reason of having the address parameters in the
INIT/INIT-ACK chunks.
2. Is it possible using the Socket API to get a
notification or anyother means of network changes
(That is if primary path fails and SCTP changes its
path automatically).
Yes it is, but I'm not sure if it is possible with the
version you are using. The socket-api of the next release
will implement the complete socket-api draft. See
http://www.ietf.org/internet-drafts/draft-ietf-tsvwg-sctpsocket-01.txt
for the details.
One of the examples will also show how to use this. Stay tuned...

3. VERY imp.. Is there a way to set the TIME the SCTP
trys to see if a path is valid before switching to the
other paths. In my case I was having 2 interfaces on
server and clients.. and the communication started and
when I break one path. THE client sent msg waits for
big time and then sends the msg. and this time
gradually reduces and after few minutes gets back to
normal speed of transfer..
I do not understand your question completely. But there
are parameters will will limit the number of retransmission
on a path before the path is declare inactive. You will be
able to adjust this parameter with next release of the socket-api.

4. If the failed path comes up again THEN why doesnt
the connection switch back to that path? or does it
need to behave that way..

Well, this is implementation dependent. If you look into RFC2960
you will find a call back API describing an "example" API for SCTP.
The procedure there says:
- if a path (primary or not) goes down, the upper layer receives a
network status change notification
- using setprimary the upper layer is able to set the primary path.
It doesn't matter if it is active or not.
So why should you switch back, why not? Does it matter?

Best regards
Michael

Regards
Dilip

=====
Live For Today & not for Tomorrow...
As Tomorrow NEVER COMES..
Do Visit My Site at http://www.angelfire.com/in/dilris

__________________________________________________
Do You Yahoo!?
Make a great connection at Yahoo! Personals.
http://personals.yahoo.com
_______________________________________________
discussion mailing list
discussion AT sctp.de
http://www.sctp.de/mailman/listinfo/discussion

Michael.Tuexen AT micmac.franken.de





Archive powered by MHonArc 2.6.19+.

Top of page