Error counters - 2620-48 Switches

[DTK100]

Hello everyone.  I'll admit immediately that i'm not that knowledgable about switches, usually work more with firewalls..

I've taken over a new (to me) environment:

• One local segment /23 - no vlans
• (6) 2620-48 switches - ports 1-48 connect to client workstations (100Mbit), port 49 connects to a "top of rack" 2910al-24G (1Gbit).  The ToR switch then connects to the firewall (1Gbit) and thus the rest of the world
• The firewall interface counters are clean
• The ToR interface counters are clean
• But, all 6 of the 2620-48 counters looks similar to the attachment.  AFAIK, no users are complaining about connectivity issues, but the enormous amount of errors has me worried.  I don't even know where to start the troubleshooting for something like this, could someone please point me in the right direction.

Many Thanks.

Dave

[johnk3r]

Hello.
Strange ToR and the firewall does not have any record / counter.

Problems with DROP TX can have several causes ...

Problem with duplex ...
Problems related to loop ...
Problems related to broadcast/multicast ...

Is there any specific configuration on the switches?

[DTK100]

john - Thanks for replying.  The config is pretty cookie cutter amungst the client connecting switches, and they look pretty plain vanilla:

Startup configuration: 10

; J9626A Configuration Editor; Created on release #RA.15.16.0005
; Ver #06:04.9c.63.ff.37.2f:1a
hostname "Building M"
max-vlans 50
timesync sntp
sntp unicast
sntp server priority 1 192.168.2.5
no stack
no telnet-server
time timezone -480
no web-management
ip default-gateway 192.168.25.1
snmp-server community "net" operator
vlan 1
name "DEFAULT_VLAN"
untagged 1-52
ip address 192.168.25.5 255.255.254.0
exit
spanning-tree
no tftp server
loop-protect 1-48
no dhcp config-file-update
password manager

[Vince-Whirlwind]

Something on your ToR switch is running at 1Gb and sending traffic to the 100Mb ports, resulting in drops.

The traffic might be traffic those ports have requested, or it could be broadcast traffic, or it could be traffic being sent to an unknown MAC address and being flooded.

Traffic sent to an unknown MAC address happens with misconfigured server interfaces.

[DTK100]

I cleared the statistics on one of these client connecting 2620s - wanting to see if i can find a pattern to when these errors start accumulating.

Almost immediately , but only on 3 ports, the "Drops Tx" started to climb - and all 3 ports climbed to the pretty much the same quantity of errors.  When i look specifically at one of these ports, there is only transmit traffic (show counters attached).  Does this seem weird?

[Vince-Whirlwind]

No, that looks exactly like what you would expect to see if a device running at 1Gb was multicasting, broadcasting or flooding to a group of ports that are only running at 100Mb.

The port can only accept less than 100Mb, the rest of the stream is recorded as dropped.