[Xorp-users] Very simple multicast setup, yet can't find any text on how to do it!

[Pavlin Radoslavov]

Erik Slagter <erik at slagter.name> wrote:

> It seems to work sometimes, but not systematic. Also lots of errors and
> warnings on stderr. Can you please have a look?
> 
> boot.config: ==========================================================

<DEL>

Config seems fine.

> ip -6 addr: ===========================================================

<DEL>

> output of xorp: (when xorp has been started, I first start the server
> application on the server and then the client application on the client on
> eth10, then I quit all of them. Multicast address is ff05::1
> 
> BTW If I can start xorp in a mode that gives less verbose but still useful
> information for you, please let me know.
> 
> =======================================================
> 
> artemis root:~erik/src/xorp/xorp-1.6/rtrmgr $ ./xorp_rtrmgr -b
> ~erik/config.boot
> [ 2009/01/18 14:27:35  INFO xorp_rtrmgr:18971 RTRMGR +249 master_conf_tree.cc
> execute ] Changed modules: interfaces, firewall, fea, mfea6, mld, rib,
> fib2mrib, pimsm6
> [ 2009/01/18 14:27:35  INFO xorp_rtrmgr:18971 RTRMGR +101 module_manager.cc
> execute ] Executing module: interfaces (fea/xorp_fea)
> [ 2009/01/18 14:27:36 INFO xorp_fea MFEA ] MFEA enabled

<DEL>

> [ 2009/01/18 14:27:56  INFO xorp_rtrmgr:18971 RTRMGR +2233 task.cc run_task ]
> No more tasks to run
> [ 2009/01/18 14:27:56 TRACE xorp_mld MLD6IGMP ] RX MLD_LISTENER_REPORT from
> fe80::20c:76ff:fe9b:3003 to ff02::1:ff80:82 on vif eth21
> [ 2009/01/18 14:27:56 TRACE xorp_mld MLD6IGMP ] Notify routing add membership
> for (::, ff02::1:ff80:82) on vif eth21
> [ 2009/01/18 14:27:56 TRACE xorp_fea MFEA ] RX kernel signal: message_type = 0
> vif_index = 0 src = :: dst = ::
> [ 2009/01/18 14:27:56 WARNING xorp_pimsm6 PIM ] RX unknown signal from MFEA_6:
> vif_index = 0 src = :: dst = :: message_type = 0
> [ 2009/01/18 14:27:58  ERROR xorp_fea:18972 FEA +1716 io_ip_socket.cc
> proto_socket_read ] proto_socket_read() failed: invalid interface pif_index
> from fe80::217:31ff:febb:3b14 to ff02::d: 0
> [ 2009/01/18 14:27:58  ERROR xorp_fea:18972 FEA +1716 io_ip_socket.cc
> proto_socket_read ] proto_socket_read() failed: invalid interface pif_index
> from fe80::204:23ff:febb:5aa9 to ff02::d: 0
> [ 2009/01/18 14:27:58 TRACE xorp_fea MFEA ] RX kernel signal: message_type = 0
> vif_index = 0 src = :: dst = ::
> [ 2009/01/18 14:27:58 WARNING xorp_pimsm6 PIM ] RX unknown signal from MFEA_6:
> vif_index = 0 src = :: dst = :: message_type = 0
> [ 2009/01/18 14:27:58 TRACE xorp_mld MLD6IGMP ] RX MLD_type_unknown from
> fe80::200:5aff:fe00:60b to fe80::204:23ff:feaa:a983 on vif eth10
> [ 2009/01/18 14:28:03 TRACE xorp_mld MLD6IGMP ] RX MLD_type_unknown from
> fe80::200:5aff:fe00:60b to fe80::204:23ff:feaa:a983 on vif eth10
> [ 2009/01/18 14:28:09  ERROR xorp_fea:18972 FEA +1716 io_ip_socket.cc
> proto_socket_read ] proto_socket_read() failed: invalid interface pif_index
> from fe80::204:23ff:febb:5aa9 to ff02::1: 0

It seems the kernel upcalls contain some bogus information.

The following email suggests that other folks might also had seen issues
with kernel upcalls:

http://www.linux-ipv6.org/ml/usagi-users/msg04077.html

Unfortunately I wasn't able to replicate the problem and investigate
the upcall issues.
In my testing I wasn't even getting the MLD_LISTENER_REPORT
delivered to the FEA, and there were no upcalls, but I didn't
investigate further. A potential source of the problem is the one
described in the following email:

http://www.linux-ipv6.org/ml/usagi-users/msg04031.html

Pavlin