Re: [openss7] Multi-Homing

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

Chuck,

Actually, there is nothing wrong with what you describe.
It is quite normal for a multi-homed host with two
interfaces which both have routes to a destination to
only use one interface for all transmitted messages.

If you really want, for example, all messages to .10.33
to go out the .1.33 interface then you have to play with
static routes or bind the socket to an interface instead
of an IP address.

So, that behavior is quite normal.  But, of course,
locking up is not.  If the lockup is only after, say,
pulling a jack to simulate failure, then the best
starting point would be sctp_route_output() in sctp_output.c.
This is the code which determines which destination address
to transmit to for multihoming.

--Brian


On Wed, 18 Jul 2001, Chuck Winters wrote:

> Brian,

> Here is a better explanation of what is going on.

> The setup is 2 machines back to back connected with crossover
> cables.  Each machine has 2 ethernet cards.  I am running a reader
> program on machine R(192.168.1.33, 192.168.10.33) and a writer
> program on machine W(192.168.1.34, 192.168.10.34). R & W are both
> set to explicitly bind to both of their addresses. 

> 	Now I ran 2 scenarios.  

> 1)  The order of the addresses sent to bind in R was .1.33,
>     .10.33.  The order of the addresses sent to bind on W was
>     .1.34, .10.34.  I started up 2 ethereal's.  One on eth0(.1
>     subnet) and one on eth1(.10 subnet).  W was setup to send to
>     .10.33(eth1).  Ethereal dumps showed that SACK, HEARTBEAT,
>     HEARTBEAT_ACK were all transmitted over the .1(eth0)
>     interface.  The DATA chunks were transmitted over the
>     .10(eth1) interface.  

> 2)  The setup was the same, except this time W was setup to send
>     to the .1(eth0) interface.  Ethereal dumps showed that all
>     HEARTBEAT, HEARTBEAT_ACK, SACK, and DATA chunks were sent over
>     the .1(eth0) interface and no traffic was observed over the
>     .10(eth1) interface.
> 
> This may help out a little with the problem.  
> 
> Chuck
>  
> On Wed, Jul 18, 2001 at 03:12:52PM -0400, Chuck Winters wrote:
> > Brian,
> > 	I had done some tests with the multi-homing capabilities.  What I found is that
> > there are no heartbeats going over the other interfaces, and after the primary
> > interface is down for a few SACK's, the receiving host locks up.  
> > 	I would like to help track this bug down.  I just need some help on where to start.
> > 
> > Thanks,
> > Chuck
> > 
> > -- 
> > Chuck Winters                            | Email:  cwinters@atl.lmco.com
> > Distributed Processing Laboratory        | Phone:  856-338-3987
> > Lockheed Martin Advanced Technology Labs |
> > 1 Federal St - A&E-3W                    |
> > Camden, NJ 08102                         |
> 
> -- 
> Chuck Winters                            | Email:  cwinters@atl.lmco.com
> Distributed Processing Laboratory        | Phone:  856-338-3987
> Lockheed Martin Advanced Technology Labs |
> 1 Federal St - A&E-3W                    |
> Camden, NJ 08102                         |

-- 
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 ¦