[f-nsp] Multicast causing high CPU

[Alexey Kouznetsov]

Here the expamle with minimum for L2/L4 pool 1 index range. Other slosts
have higher values. All other ranges always bigger 1024 for all modules


==== SLOT 6 CAM PARTITION ====
DMA: 20 (0x14)
Number of CAM devices per DMA:  8
Number of hw entries per CAM:  0x00800
Total size of CAM = 1Mbits
complete CAM index range per DMA:
  (sw) 1 - 16383  (1 - 0x03fff), total entries: 16383 (0x03fff)
  (hw) 0 - 16383  (0 - 0x03fff), total entries: 16384 (0x04000)
Percentage of CAM hardware entries for each partition:
  Layer3 = 10239 (0.624938Mbits)        (62.493896%)
     Level3 = 2047 (0.124938Mbits)      (12.493896%)
     Level2 = 2048 (0.125Mbits) (12.5%)
  Layer4 Pool0 = 2048 (0.125Mbits)      (12.5%)
  Layer2/Layer4 Pool1,2,3 = 4096 (0.25Mbits)    (25%)
L3 level 3 index range:
  (sw) 1 - 2047         (0x00001 - 0x007ff), free 2046 (0x007fe)
  (hw) 1 - 2047         (0x00001 - 0x007ff)
L3 level 2 index range:
  (sw) 2048 - 4095      (0x00800 - 0x00fff), free 2048 (0x00800)
  (hw) 2048 - 4095      (0x00800 - 0x00fff)
L3 index range:
  (sw) 4096 - 10239     (0x01000 - 0x027ff), free 5899 (0x0170b)
  (hw) 4096 - 10239     (0x01000 - 0x027ff)
L4 pool 0 index range:
  (sw) 10240 - 12287    (0x02800 - 0x02fff), free 2044 (0x007fc)
  (hw) 10240 - 12287    (0x02800 - 0x02fff)
L2/L4 pool 1 index range:
  (sw) 12288 - 14335    (0x03000 - 0x037ff), free 568 (0x00238)
  (hw) 12288 - 14335    (0x03000 - 0x037ff)
L2/L4 pool 2 index range:
  (sw) 14336 - 16383    (0x03800 - 0x03fff), free 1613 (0x0064d)
  (hw) 14336 - 16383    (0x03800 - 0x03fff)
DMA: 22 (0x16)
Number of CAM devices per DMA:  8
Number of hw entries per CAM:  0x00800
Total size of CAM = 1Mbits
complete CAM index range per DMA:
  (sw) 1 - 16383  (1 - 0x03fff), total entries: 16383 (0x03fff)
  (hw) 0 - 16383  (0 - 0x03fff), total entries: 16384 (0x04000)
Percentage of CAM hardware entries for each partition:
  Layer3 = 10239 (0.624938Mbits)        (62.493896%)
     Level3 = 2047 (0.124938Mbits)      (12.493896%)
     Level2 = 2048 (0.125Mbits) (12.5%)
  Layer4 Pool0 = 2048 (0.125Mbits)      (12.5%)
  Layer2/Layer4 Pool1,2,3 = 4096 (0.25Mbits)    (25%)
L3 level 3 index range:
  (sw) 1 - 2047         (0x00001 - 0x007ff), free 2046 (0x007fe)
  (hw) 1 - 2047         (0x00001 - 0x007ff)
L3 level 2 index range:
  (sw) 2048 - 4095      (0x00800 - 0x00fff), free 2048 (0x00800)
  (hw) 2048 - 4095      (0x00800 - 0x00fff)
L3 index range:
  (sw) 4096 - 10239     (0x01000 - 0x027ff), free 5899 (0x0170b)
  (hw) 4096 - 10239     (0x01000 - 0x027ff)
L4 pool 0 index range:
  (sw) 10240 - 12287    (0x02800 - 0x02fff), free 2044 (0x007fc)
  (hw) 10240 - 12287    (0x02800 - 0x02fff)
L2/L4 pool 1 index range:
  (sw) 12288 - 14335    (0x03000 - 0x037ff), free 1765 (0x006e5)
  (hw) 12288 - 14335    (0x03000 - 0x037ff)
L2/L4 pool 2 index range:
  (sw) 14336 - 16383    (0x03800 - 0x03fff), free 2042 (0x007fa)
  (hw) 14336 - 16383    (0x03800 - 0x03fff)



2009/3/16 Jeroen Wunnink <jeroen at easyhosting.nl>

> Do a: sh cam-partition detail
>
> See if any of the cam counters are (near) zero, if so you might need to
> restructure your cam partitions. When running out of CAM, the packets will
> be processed over CPU, something you really don't want to do on a bigiron.
>
> Alexey Kouznetsov wrote:
>
>> There are totally no IP ROUTING. Only MCAST ROUTING and L2 switching. Now
>> I have added no icmp redirects. but as far as here are no IP routing
>> traffic, there are no icmp redirect generates and no any CPU changes seen.
>> Also I already have
>>  no ip source-route
>> in the config
>> There are no RIP, no OSPF. only MBGP routing protocol.
>> # egrep '^router' config
>> router dvmrp
>> router pim
>> router msdp
>> router pim
>> router bgp
>>  in bgp router no any unicast active neighbors. only for address-family
>> ipv4 multicast
>> we already have
>> ip multicast-perf
>> in our config for at list last year. (switch was rebooted after this
>> command added)
>>  Same as we already added
>>  mcast-hw-replic-oar
>> ip multicast hardware-drop
>> in config and it was reboted after such commands added. No any changes
>> seen.
>>  Also I tried to add
>> router pim
>> + hardware-drop
>> and no any changes seen.
>>
>> /Alexey
>> 2009/3/12 Dimitar Kosadinov <kgb at bginfo.net <mailto:kgb at bginfo.net>>
>>
>>    always use "no ip icmp redirect" !!!
>>    this safe Your CPU :)
>>    I dont sure Your multicast problem is this, but just use it. This
>>    reflect to OLD RIPv1 based routing only, I sure you dont use rip.
>>    try use "ip multicast-perf"  too / need reload btw to see changes /
>>    sorry for my bad english
>>
>>
> --
>
> Met vriendelijke groet,
>
> Jeroen Wunnink,
> EasyHosting B.V. Systeembeheerder
> systeembeheer at easyhosting.nl
>
> telefoon:+31 (035) 6285455              Postbus 48
> fax: +31 (035) 6838242                  3755 ZG Eemnes
>
> http://www.easyhosting.nl
> http://www.easycolocate.nl
>
>
> _______________________________________________
> foundry-nsp mailing list
> foundry-nsp at puck.nether.net
> http://puck.nether.net/mailman/listinfo/foundry-nsp
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/foundry-nsp/attachments/20090317/e10a312c/attachment.html>