[hipl-users] Re: Mobilty test
- From: Miika Komu <miika@xxxxxx>
- To: hipl-users <hipl-users@xxxxxxxxxxxxx>
- Date: Sun, 5 Aug 2007 11:34:36 +0300 (EEST)
On Thu, 2 Aug 2007, Fco. Javier Melero wrote:
Hi Javier,
2.- Roam from A to B and return to A works almost fine. Sometimes it is
necessary to delete the old ip for the MN to trigger update location
exchange. And sometimes the update is sended to the old router, so I
must wait until that route is dismissed (about 10 s.) and then delete
the new ip, which force it to autoconfigure again and send a new update
location exchange.
3.- Roam from A to C works fine and quicker than any other case (not
necessary to delete old ip). But then, going outside of C never works.
No matter what I do, the update location exchange is never triggered.
And my questions are:
Is it possible to get a smoother roaming with current hipl
implementation? Otherwise, I guest I should do my tests in the old
fashioned way (both nodes on the same network and changing ip manually).
We haven't tested the mobility code extensively with different DHCP/RARVD
configurations, but we'd be happy to look at the problem. Could you give
us more feedback? Particularly, what does the hipd say when the handover
should occur?
If you are familiar with gdb, you can do also the following:
gdb hipd
br hip_netdev_event
run
The hipd/netdev.c:hip_netdev_event() function catches all route etc.
information from the kernel. Particularly interesting here is the
RTM_NEWADDR / RTM_DELADDR case.
Do you have also IPv4 addresses in your networks? Interfamily handovers
are in the work queue currently. One reason for failure might be that the
hipd marks an IPv4 address as active and we haven't implemented yet
interfamily (from IPv6 to IPv4 and vice versa) handovers yet. This is in
the work queue though.
What is the "hipconf handoff mode" option for? I have tried both, lazy
and active, and I haven´t notice any difference.
It is related how the hipd handles multiple simultaneous addresses. In
active mode, the hipd should switch immediately to use the latest address
that was added. Lazy mode is the opposite of this; it changes the current
active address only when the previous address was deleted.
This is a new feature and it has not been tested throughly. It might have
bugs and we'd be happy to get your feedback on it.
--
Miika Komu http://www.iki.fi/miika/
Other related posts:
