[c-nsp] Cisco 3850-24XS Output Discards, Buffer Sizes of TenGig interfaces

Giles Coochey giles at coochey.net
Fri Oct 12 08:54:58 EDT 2018


The bug on the cisco site appears to be:
CSCvb89733

We've implemented :

qos queue-softmax-multiplier 1200

And this has alleviated  the drops for the time being, by allocating 
more softmax, but hardmax remains at 120...

The bug says this is present in 16.3.3a, so not sure why we're getting 
it in 16.3.5b?


On 12/10/2018 12:14, Giles Coochey wrote:
> Hi All,
>
> I am managing a number of 3850-24XS Catalysts, running Denali 16.3.5b.
>
> These have 24 10Gig interfaces and they're seeing a large number of 
> output discards on these.
>
> The output of the following command shows the hardmax/softmax queues 
> on the interface:
>
> #show platform hardware fed switch active qos queue config interface 
> te1/0/1
> DATA Port:7 GPN:873 AFD:Disabled QoSMap:0 HW Queues: 56 - 63
>   DrainFast:Disabled PortSoftStart:2 - 1080
> ----------------------------------------------------------
>   DTS Hardmax   Softmax  PortSMin GlblSMin  PortStEnd
>   --- --------  -------- -------- --------- ---------
>  0   1  5   120  2   480  6   320   0     0   5  1440
>  1   1  4     0  6   720  3   480   2   180   5  1440
>  2   1  4     0  5     0  5     0   0     0   5  1440
>  3   1  4     0  5     0  5     0   0     0   5  1440
>  4   1  4     0  5     0  5     0   0     0   5  1440
>  5   1  4     0  5     0  5     0   0     0   5  1440
>  6   1  4     0  5     0  5     0   0     0   5  1440
>  7   1  4     0  5     0  5     0   0     0   5  1440
>
> According to the following link about Troubleshooting Output drops on 
> this platform these are the buffer values for a 1Gig interface?
>
> https://www.cisco.com/c/en/us/support/docs/switches/catalyst-3850-series-switches/200594-Catalyst-3850-Troubleshooting-Output-dr.html 
>
>
> Does anyone know whether there is a bug-id for this, and whether it is 
> fixed in any Denali 16.3.x ?
>
> In the document it specifies that:
>
> The show command that can be used to see this allocation is ‘show 
> platform qos queue config <interface>’.
>
> For a 1Gig interface,
>
> 3850#show platform qos queue config gigabitEthernet 1/0/1
> DATA Port:20 GPN:66 AFD:Disabled QoSMap:0 HW Queues: 160 - 167
>   DrainFast:Disabled PortSoftStart:1 - 1080
> ----------------------------------------------------------
>   DTS Hardmax   Softmax  PortSMin GlblSMin  PortStEnd
>   --- --------  -------- -------- --------- ---------
>  0   1  5   120  6   480  6   320   0     0   3  1440
>  1   1  4     0  7   720  3   480   2   180   3  1440
>  2   1  4     0  5     0  5     0   0     0   3  1440
>  3   1  4     0  5     0  5     0   0     0   3  1440
>  4   1  4     0  5     0  5     0   0     0   3  1440
>  5   1  4     0  5     0  5     0   0     0   3  1440
>  6   1  4     0  5     0  5     0   0     0   3  1440
>  7   1  4     0  5     0  5     0   0     0   3  1440
>
> <<output omitted>>
>
> For a 10Gig interface,
>
> 3850#show platform qos queue config tenGigabitEthernet 1/0/37
> DATA Port:1 GPN:37 AFD:Disabled QoSMap:0 HW Queues: 8 - 15
>   DrainFast:Disabled PortSoftStart:2 - 6480
> ----------------------------------------------------------
>   DTS Hardmax   Softmax  PortSMin GlblSMin  PortStEnd
>   --- --------  -------- -------- --------- ---------
>  0   1  6   720  8  2880  7  1280   0     0   4  8640
>  1   1  4     0  9  4320  8  1920   3  1080   4  8640
>  2   1  4     0  5     0  5     0   0     0   4  8640
>  3   1  4     0  5     0  5     0   0     0   4  8640
>  4   1  4     0  5     0  5     0   0     0   4  8640
>
> _______________________________________________
> 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