[f-nsp] BigIron MG8 hight CPU load on LP card

[Aaron]

You should be able to log into the LP and do a debug packet cap. That 
will show you what packets are hitting the LP cpu.


On 2/24/2014 9:48 AM, sergp at soho.net.ua wrote:
> Hallo
> We have the router :
> HW: BigIron MG8 Router
> SL M1: BIMG8-MR Management Module
> Boot: Version 2.3.0jT105 Copyright © 1996-2003 Foundry Networks, Inc.
> Monitor: Version 2.3.0jT105 Copyright © 1996-2003 Foundry Networks, Inc.
> IronWare: Version 2.3.0jT103 Copyright © 1996-2003 Foundry Networks, Inc.
>
>         Module Status Ports Starting MAC
> M1 (upper): BIMG8-MR Management Module Active
> M2 (lower): BIMG8-MR Management Module Standby (Ready State)
> F1: BIMG8-SFM Switch Fabric Module Active
> S1: BIMG8-1Gx40-SFP-v6 40 -port 1 GbE IPv4/v6 Module (IPv6 + OACL) 
> CARD_STATE_UP 40
> S5: BIMG8-10Gx8-v6 8 -port 10GbE IPv4/v6 Module (IPv6 + OACL) 
> CARD_STATE_UP 8
> S8: BIMG8-10Gx4-v6-A 4 -port 10GbE IPv4/v6 Module (IPv6 + OACL) 
> CARD_STATE_UP 4
>
> We have 5 BGP uplinks come through two 10Gb ports 5 and 8 modules. and 
> five switches with servers connected 1Gb ports in 1 module .
>
> When the total traffic of up to 1 gigabit but a relatively large 
> number of small packets 400-600 Kpps cpu first module is loaded at 
> 100% and start losing 80-90%.
> while CPU on the control module and 10G modules in slots 5 and 8 are 
> not loaded .
>
> it looks like this:
> show cpu lp
> SLOT #: LP CPU UTILIZATION in%:
>                in 1 second: in 5 seconds: in 60 seconds: in 300 seconds:
>   1                 96                     96 100                    100
>   5                  0                        0 1                   
>      1
>   8                  0                        0 0               
>          0
>
> show cpu
> ... Usage average for all tasks in the last 1 seconds ...
> ================================================== ========
> Name         us /sec            %
> idle         665002             100
> ...
>
> if the number of packets reaches 1Mpps already traffic through 1 
> module does not go
> and recovered only work when disconnected stream of small packets .
>
> May be someone faced with a similar problem or have ideas what can i do?
> _______________________________________________
> foundry-nsp mailing list
> foundry-nsp at puck.nether.net
> http://puck.nether.net/mailman/listinfo/foundry-nsp
>