On Mon, Nov 06, 2006 at 06:32:49AM -0500, Daniel Ouellet wrote:
> Daniel Ouellet wrote:
> >Hi,
> >
> >I am trying to isolate this issue, but the exact same setup and
> >configuration for the 3.9 was working and after the upgrades to 4.0
> >without any changes what so ever to the bgpd.conf doesn't work anymore.
> >
> >All bgp sessions are up as before, all ibgp sessions are up as well, but
> >all routers that were configure to use this bgpd as a route-reflector do
> >not get the full table anymore, but only the ibgp routers.
> >
> >Looking in the man page, I don't see a new option that may have impacted
> >this, or that should be added to enable it again.
> >
> >Any clue as to where I should look, or if anything changed between the
> >3.9 and 4.0 in regards to route-reflector setup.
> >
> >So, far I find nothing in this regards.
>
> Here is more details to show the problem.
>
> It should really announce the full routing tables based on the
> configuration below as an example, but it doesn't.
>
> I use real full feeds from a few real peers to show real data and did a
> configuration for testing with the following results below.
>
> Why wouldn't it send the full table as a reflector would?
>
> Any clue as to what I forgot in the new version oppose to 3.9, or is
> this really a bug then? I can't get the reflector to work anymore in 4.0.
>
> Thanks
>
> ============================================
>
> # bgpctl sh rib memory
> RDE memory statistics
> 198197 IPv4 network entries using 12.1M of memory
> 227458 prefix entries using 12.1M of memory
> 43193 BGP path attribute entries using 4.9M of memory
> 38816 BGP AS-PATH attribute entries using 1.5M of memory,
> and holding 43193 references
> 3814 BGP attributes entries using 149K of memory
> and holding 138391 references
> 3813 BGP attributes using 22.3K of memory
> RIB using 30.9M of memory
>
> ================================================
>
> Showing the count of valid announce routes, only 81 learn from that
> local router bgp sessions
>
> # bgpctl show rib neighbor x.x.x.192 out | grep -c '*>'
> 81
>
> ================================================
>
> Simple test configuration:
>
> # global configuration
> AS xxxxx
> router-id x.x.x.199
>
> # Network announcements
> network x.x.x.199/32
>
> neighbor x.x.x.192 {
> remote-as xxxxx
> descr "Test"
> multihop 5
> local-address x.x.x.199
> holdtime 180
> holdtime min 3
> route-reflector
> announce all
> tcp md5sig password TestOnly
> }
>
> <snip>
>
> ==================================================
>
> # bgpctl s s
> Neighbor AS MsgRcvd MsgSent OutQ Up/Down State/PrefixRcvd
> Test xxxxx 260 336 0 00:04:38 57
> <snip>
>
Please check that the routes on your route-reflector. My guess is that you
need to set "nexthop qualify via bgp" at least that was the error I had
while testing it now. Afterwards route reflection worked for me.
--
:wq Claudio
