[c-nsp] Long list of route-maps

Arie Vayner (avayner) avayner at cisco.com
Thu Mar 11 13:02:21 EST 2010


This is true, but the grouping of the policies should be for the egress
policies, while the ingress policies can stay the same.
Also, as this is a peering point, I do not think that whatever is
received should be sent to the other peers, so I would say that the 1st
policy entry on any outgoing policy should be to block this kind of
advertisements (set a community for a peer on ingress, and block to
other peers).

I think increasing the MTU (PMTUD) would help as it would reduce the
number of packets.

If nothing helps, it could be a good idea to try a COPP policy which
polices the incoming BGP updates to a decent value so at link up events
you pace the updates, but do not overload the CPU, which would end up
with a faster convergence...

Another point to make sure is that you do not overload the TCAM table by
installing too many routes. If it's full, traffic would start hitting
the CPU... (I do not think this is the case here)

Arie

-----Original Message-----
From: Sven Huster [mailto:sven at huster.me.uk] 
Sent: Thursday, March 11, 2010 19:46
To: Arie Vayner (avayner)
Cc: Paul Stewart; Andy B.; cisco-nsp at puck.nether.net
Subject: Re: [c-nsp] Long list of route-maps

If any of the received updates lead to a new best path surely these
updates get processed by the 150 outbound route-maps now

The statement the grouping in update groups doesn't help is not
necessarily true here.

--
Sven

On 11 Mar 2010, at 17:33, Arie Vayner (avayner) wrote:

> This would not actually help much as still each received update has to
> be analyzed separately.
> The grouping is important for egress policies - all BGP peers with the
> same egress policy would be placed into the same BGP update group
> dramatically reducing processing of outgoing updates.
> 
> Arie
> 
> -----Original Message-----
> From: cisco-nsp-bounces at puck.nether.net
> [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Paul Stewart
> Sent: Thursday, March 11, 2010 19:27
> To: 'Andy B.'; cisco-nsp at puck.nether.net
> Subject: Re: [c-nsp] Long list of route-maps
> 
> Why a route-map PER peer?  Can you not group them under the same
> conditions
> and simplify things a bit?
> 
> This may not be the problem .... sounds like something else possibly..
> 
> Paul
> 
> 
> -----Original Message-----
> From: cisco-nsp-bounces at puck.nether.net
> [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Andy B.
> Sent: Thursday, March 11, 2010 12:19 PM
> To: cisco-nsp at puck.nether.net
> Subject: [c-nsp] Long list of route-maps
> 
> I feel desperate: I just turned up a new Transit Session with an
> upstream and my router goes nuts and is dropping other BGP sessions on
> it: 4/0 (hold time expired) 0 bytes
> 
> The situation is like this:
> 
> The router is peering on a public IX with approximatively 150 members.
> Each BGP session has its own route-map, so the list is really BIG!
> 
> When I turned up my transit about an hour ago, CPU went to 100% and is
> still at 100% right now and it drops BGP peers and brings them back,
> and drops and brings them back, ... I'm in a loop and I think the only
> way to get out of that look is to bring up each bgp peer step by step
> - really not an option.
> 
> CPU utilization for five seconds: 100%/5%; one minute: 99%; five
> minutes:
> 99%
> PID Runtime(ms)   Invoked      uSecs   5Sec   1Min   5Min TTY Process
> 442    56982884  32932073       1730 83.93% 85.54% 82.20%   0 BGP
> Router
> 329     1639012   1857164        882  3.35%  2.01%  3.28%   0 IP RIB
> Update
> 403     6686764   2462837       2715  1.91%  0.71%  0.81%   0 BGP
> Scheduler
> 273     7514324  63409992        118  1.51%  1.55%  1.44%   0 IP Input
> 340      421908   2376861        177  1.35%  0.63%  0.96%   0 XDR
mcast
> 553     3487144  30648800        113  0.87%  1.21%  1.28%   0 BGP I/O
>   9    35017808   1752692      19979  0.79%  0.54%  0.50%   0 Check
> heaps
> 550     7132896  84539324         84  0.47%  0.30%  0.32%   0 IPv6
> Input
>  12    20351908 188141150        108  0.15%  0.54%  0.36%   0 ARP
Input
> 493      465108   4375354        106  0.07%  0.05%  0.03%   0 Port
> manager
> per
>  66        7916    202448         39  0.07%  0.00%  0.00%   0 BGP Open
> 333      284012  26233730         10  0.07%  0.16%  0.15%   0 TCP
Timer
> 402       41152  73291500          0  0.07%  0.00%  0.00%   0 RADIUS
>  51      155388   2479245         62  0.07%  0.05%  0.05%   0
> Per-Second
> Jobs
>  95       45504   2448923         18  0.07%  0.00%  0.00%   0
Heartbeat
> Proces
>  24     9277560  71169033        130  0.00%  0.10%  0.11%   0 IPC Seat
> Manager
>  52     1725428     43152      39984  0.00%  0.08%  0.05%   0
> Per-minute
> Jobs
> 328     5111328     41990     121727  0.00%  0.23%  0.18%   0 IP
> Background
> 341      723640    487509       1484  0.00%  0.01%  0.00%   0 IPC LC
> Message H
> 353      851584   3607096        236  0.00%  0.03%  0.04%   0 CEF:
IPv4
> proces
> 371       20720   2473579          8  0.00%  0.01%  0.00%   0 OSPF-1
> Router
> 372      475340   1244448        381  0.00%  0.03%  0.02%   0 HIDDEN
> VLAN
> Proc
> 546       33060    117785        280  0.00%  0.00%  0.03%   0 IPv6 RIB
> Redistr
> 551       77312  12335517          6  0.00%  0.03%  0.04%   0 IPv6 ND
> 560       49732     17676       2813  0.00%  0.01%  0.10%   0 SNMP
> Traps
> 562       41720       903      46201  0.00%  0.00%  0.23%   0
> Collection
> proce
> 563    55952820    420289     133129  0.00%  0.88%  1.80%   0 BGP
> Scanner
> 565         784       200       3920  0.00%  0.15%  0.13%   1 SSH
> Process
> 
> 
> 6500 box with SXI3
> 
> What is eating my router's CPU?
> Is it the big list of route-maps?
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
> 
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
> 
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/




More information about the cisco-nsp mailing list