[Olsr-dev] strange log message when received netlink error

[Henning Rogge]

Okay, that explains a few of the problems.

I would suggest we add some more filters in the HNA and TC input code.
We must prevent malformed data to get into the OLSR database from
other nodes. If a message contains such data, we should just drop it.

What do you think?

Henning Rogge

On Fri, Mar 29, 2013 at 1:26 AM, Saverio Proto <(spam-protected)> wrote:
> Hello,
>
> I am back to the topic:
> Received netlink error code Invalid argument (-22)
>
> I tracked down the problem. The fact is that in our network we are
> testing plugins that are unstable, so we have unstable routers pushing
> impossible routes to neighbors. We have networks with '0'netmask, or
> with '0' as a gateway. Wow !
>
> the point is that the current code is not handling the case were an
> impossible route is pushed to the kernel with a netlink socket, and
> the kernel replies -22.
>
> Until the buggy routers are online we get messages like:
>
> Mar 26 23:43:30 svn olsrd[28208]: Add route 192.168.178.0/0 from
> 192.168.178.2 via 10.0.1.177, cost 32.361, metric 6, v 525918:
> Resource temporarily unavailable
> Mar 26 23:43:30 svn olsrd[28208]: Received netlink error code Invalid
> argument (-22)
>
> When the buggy routers go offline, the route will expire and we get
> the same message forver but this time is a "Del route" message.
>
> So when the route expires we get infinites retries of deleting this
> route from the kernel because the route exists in the olsrd process
> database.
>
> In src/process_routes.c line 343 we delete the route from the
> datastructure only of deletion from Kernel was successfull.
>
> Where is the right point of the code to handle this situation ?
>
> Saverio



-- 
We began as wanderers, and we are wanderers still. We have lingured
long enough on the shores of the cosmic ocean. We are ready at last to
set sail for the stars - Carl Sagan