[Xorp-users] BGP router is not receiving BGP routes
Edwin Schokkenbroek
edwin at spacebugs.nl
Fri Aug 20 16:01:52 PDT 2010
On Aug 21, 2010, at 12:33 AM, Ben Greear wrote:
> On 08/20/2010 03:25 PM, Edwin Schokkenbroek wrote:
>>
>> On Aug 19, 2010, at 8:36 PM, Aleksandar Cvjetic wrote:
>>
>>> Hello,
>>>
>>> On R4 try to configure a policy matching protocol OSPF routes and apply it as export BGP policy (as you did for static routes on R13). I think that Cisco logic doesn't work here, at least you don't have an option for "network" statement to advertise routes into BGP.
>>
>> Thanks for pointing this out. And Well Cisco logic, doesn't work indeed. After fighting with bgp and policies for a day and a long night, I got it working. I'm quite surprised to find out that only a policy which says: from protocol ospf ,for example isn't going to work. A network statement is required, which tells what prefix needs to be announced. In Junos I would do something like: set term 1 from protocol static ; set term 1 then accept. And 'm done (basically). Now that I begin to understand how it works with Xorp, it provides a very clean way to have full control over what prefix is to be announced by bgp.
>
> By the way, there were complaints from years ago that xorp bgp has performance issues when
> synchronizing with large route databases (I think the users were connecting to the full internet
> BGP routing table, or something like that).
>
> I was never able to set up a test to try to debug/optimize this, so if you
> are able to set up such a thing, I'm interested in knowing the results.
Should not be a problem. Right now I'm running multiple Xorp instances in jails, which reside in a virtual machine :
JID IP Address Hostname Path
1 - local /
2 - R4 /
3 - R10 /
4 - R5 /
5 - R6 /
6 - R13 /
7 - R15 /
8 - R9 /
9 - R11 /
10 - R12 /
11 - R1 /
12 - R2 /
13 - R3 /
14 - R14 /
I have a smal amount of prefixes, but with 15 jails running Xorp the machine load and memory usage is not high:
last pid: 13948; load averages: 0.01, 0.02, 0.00 up 1+01:36:46 00:54:49
118 processes: 1 running, 117 sleeping
CPU: 0.0% user, 0.0% nice, 0.4% system, 0.0% interrupt, 99.6% idle
Mem: 288M Active, 56M Inact, 136M Wired, 111M Buf, 499M Free
Swap: 256M Total, 256M Free
When dealing with full routing tables ... I can image that a couple of these instances can effect performance. What is the best test to do ?
Testing in a virtual machine, or testing it on real hardware (or both)? Memory will become a serious issue when loading full BGP routing table(s).
Edwin
>
> Thanks,
> Ben
>
> --
> Ben Greear <greearb at candelatech.com>
> Candela Technologies Inc http://www.candelatech.com
>
More information about the Xorp-users
mailing list