High In Discards on L3 interface with S8

[Anonymous]

Hi,

Investigating an issue where some packet loss is happening within the local network.

Looking into where packets maybe being dropped there is a continual high in discard count on a L3 interface between a VSP and S8. In fact, it is happing on L2 connection too, but focusing on the L3 interface that just has the one tagged VLAN helps with the focus.

Looking at both sides, the S8 is showing no drop packets at all and VSP is also clean on out discards by looking at the following:

https://extremeportal.force.com/ExtrArticleDetail?an=000080052&q=vsp%20in%20discards

The discards shown below keep incrementing by maybe about 1000 or more just cycling the command below immediately after each other:

CR1:1(config)#% show interfaces gigabitEthernet statistics 1/23
************************************************************************************
Command Execution Time: Sat Oct 01 07:52:44 2022 GMT
************************************************************************************
Please widen the terminal for optimal viewing of data.
==================================================================================================================
Port Stats Interface
==================================================================================================================
PORT IN OUT IN OUT
NUM OCTETS OCTETS PACKET PACKET
------------------------------------------------------------------------------------------------------------------
1/23 81798737470 7320887345 65579668 25187527

PORT IN OUT IN OUT OUTLOSS
NUM FLOWCTRL FLOWCTRL PFC PFC PACKETS
------------------------------------------------------------------------------------------------------------------
1/23 0 0 0 0 0

When I look at the port utilisation it is probably about 2-3%, so doesn't seem to be an oversubscription issue.

Looking at the following article, none of these seem to fit with the issue:

https://extremeportal.force.com/ExtrArticleDetail?an=000081496&q=in%20discards

This is the configuration on the VSP side:

interface GigabitEthernet 1/23
encapsulation dot1q
exit
vlan members remove 1 1/1-1/24,2/1-2/24,3/1-3/24,4/1-4/24 portmember
vlan members 3001 1/23 portmember
interface GigabitEthernet 1/23
untagged-frames-discard
default-vlan-id 3001
no shutdown
exit

This is the configuration the S8 (EOS) side:

set linkflap portstate enable tg.3.12
set linkflap threshold tg.3.12 10
set linkflap interval tg.3.12 60
set linkflap downtime tg.3.12 14400
set port alias tg.3.12 SPARE
set spantree portadmin tg.3.12 disable
set spantree adminpathcost tg.3.12 50 sid 0
set spantree adminpathcost tg.3.12 50 sid 101
set spantree adminpathcost tg.3.12 100 sid 102
set vlan egress 3001 tg.3.12 tagged

The both have Extreme badged 10GbLR's, and show clean L1 errors (like FCS) on both sides other then the in discards on the VSP.

The VSP is a 8600 running version 8.0.4.0.GA. Which I appreciate is getting a little long in the tooth.

The S8 is running 08.62.04.0002.

It would be nice in a way if this was dropping packets that validated the packet loss seen in the network via PCAPs, but this seems to be maybe the VSP just dropping something it doesn't like, as an example STP BPDU's, although this is disabled on the port on the S8 side.

Is there any advise to figure out what exactly the in discards are?

Many thanks in advance.

[PeterK]

You have "untagged-frames-discard" on the VSP.

This could be the reason. Can you try to remove the command?

First I thought to stp. But you have disabled on S-Series.

Can you verify that flowcontrol is diabled on S8? Normaly flowcontrol could be responsible for in-discards on EOS Devices. Of course the VSP shows no pause-frames received, but I would check this.

What else have you configured on S8 Port?

Do you have "vlan dynamicegress" enabled (default) ?