nexus3548 multicast packet drop issue 2 (WOULD REALLY APPRECIATE HELP)

[johnyim12345678]

https://community.cisco.com/t5/switching/nexus3548-multicast-packet-drop-issue/td-p/4551503

Hello, I have posted a question regarding my conpany's nexus3548 switch packet drop issue yesterday on the link above, and someone thankfully has suggested me to change wrr unicast-bandwidth settings. My boss has accepted the suggestion and changed the ratio to multicast 9 : unicast 1, but it seems like the issue is still not fixed..... Our clients keep complaining and we are really panic now since none of us are an expert on network devices. Now my boss is doubting maybe the issue is caused by buffer config. In order to go through it, we have enabled ABM(Active Buffer Monitoring) and checked the buffer status, and we found out something very weird.  

SASW-SND2# show hardware profile buffer monitor summary

slot 1
=======

Summary CLI issued at: 02/15/2022 16:52:54

Maximum buffer utilization detected
1sec 5sec 60sec 5min 1hr
------ ------ ------ ------ ------
Buffer Block 1 384KB 1152KB 1536KB 3456KB 4992KB

Total Shared Buffer Available = 5397 Kbytes
Class Threshold Limit = 4965 Kbytes
============================================================
Buffer Block 2 384KB 1152KB 1536KB 3456KB 4992KB

Total Shared Buffer Available = 5799 Kbytes
Class Threshold Limit = 5349 Kbytes
============================================================
Buffer Block 3 0KB 0KB 0KB 0KB 0KB

Total Shared Buffer Available = 5799 Kbytes
Class Threshold Limit = 5349 Kbytes

First, when we check out the summary through buffer monitor summary(brief) command, it seems like buffer block 3 is not being used at all. It this supposed to be like this? Sorry for my ignorance, but none of our team members even don't know how to interprete those results. Please help me understand the result and know if it's okay for buffer block 3 to not be used at all.

Also, whenever we input this command, even when there's extremely small amount of traffic, the command output always warns that maximum buffer utilization is detected. Why do you think this is happening? When we check the detailed output using the show hardware profile buffer monitor detail command, we don't see any non-zero values on the last four columns, which means the maximum buffer is not reached actually, if we interpreted it right. 

I am very sorry to bother y'all, but pls help me interprete this buffer monitoring result and resolve this severe multicast packet drop issue! Our team desperately needs your help. Thank you!

[paul driver]

Hello

This is a duplcate post here

Also I guess you did not you review the bug notification from that OP - which highlighted the same issue you are experiencing- here

[Georg Pauwen]

Hello,

you say that 20 of your ports are in use. How are these ports physically distributed ?

The buffer blocks are mapped like this:

Buffer Block #1: Front-Panel Ports 1-16
Buffer Block #2: Front-Panel Ports 17-32
Buffer Block #3: Front-Panel Ports 33-48

If possible, make sure that the 20 ports are evenly distributed/patched over the buffer blocks. So if you use these ports:

Ports 1-7

Ports 17-24

Ports 33-40

If your connections are physically patched to these ports, that would mean that all three buffer blocks are more or less evenly used.

If you use only ports 1-32, that would mean that buffer block 3 is never used.

I hope that makes sense...