output drops

[venkat.moturi]

Seeing output drops incrementing (not rapidly but steadily) on the Gig Eth Interfaces to which the Servers are connected.

Is this merely a heavy traffic from/to the server causing the Switch to drop those packets or anything pertaining to the Server NIC which is causing the issue of delay for the Client

Cisco Catalyst 3750

 MTU 1500 bytes, BW 1000000 Kbit/sec, DLY 10 usec,
     reliability 255/255, txload 3/255, rxload 7/255
  Encapsulation ARPA, loopback not set
  Keepalive set (10 sec)
  Full-duplex, 1000Mb/s, media type is 10/100/1000BaseTX
  input flow-control is off, output flow-control is unsupported
  ARP type: ARPA, ARP Timeout 04:00:00
  Last input never, output 00:00:01, output hang never
  Last clearing of "show interface" counters 49w3d

Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 186332

Queueing strategy: fifo
  Output queue: 0/40 (size/max)
  30 second input rate 30373000 bits/sec, 6654 packets/sec
  30 second output rate 13639000 bits/sec, 4975 packets/sec
     73982548647 packets input, 44425199876947 bytes, 0 no buffer
     Received 1482191 broadcasts (74574 multicasts)
     0 runts, 0 giants, 0 throttles
     0 input errors, 0 CRC, 0 frame, 0 overrun, 0 ignored
     0 watchdog, 74574 multicast, 0 pause input
     0 input packets with dribble condition detected
     54249454875 packets output, 20936680480788 bytes, 0 underruns
     0 output errors, 0 collisions, 0 interface resets
     0 unknown protocol drops
     0 babbles, 0 late collision, 0 deferred
     0 lost carrier, 0 no carrier, 0 pause output
     0 output buffer failures, 0 output buffers swapped out

Need advise

[Mark Malone · Accepted solution]

Its most likely the 3750 they have really low buffers and are renowned for total output drops incrementing , you don't get it in the newer switches they increased the buffer from 6mb to 12mb on 3ks

The buffer is most likely filling and dropping especially when servers connected , these are really access switches there not built for server traffic but they do work fine with server too but you can expect drops

layer 2 qos can be the cause as well and can also fix it if tweaked mls

just google output drops on 3750 you will see 100s of cases on diff forums , if its not causing you an issue you could  ignore it

as well what you can check is the amount of drops against the volume of traffic pushed through if its less than 1% it should be ignored anyway as there will always be some drops due to certain types of application traffic that burst and flood the buffer ,

so there are several things that can cause it , buffer overflow , mls qos , certain app burst traffic which can be checked through wireshark

http://www.cisco.com/c/en/us/support/docs/lan-switching/switched-port-analyzer-span/116260-technote-wireshark-00.html

There is no real quick fix you will need to narrow down whats causing it exactly then implement the fix whether qos

[venkat.moturi]

Thanks for taking time to answer my query.

Thank you very much, it helps

[Joseph W. Doherty]

As Mark as already noted, the 3750s are a bit infamous for dropping packets due to their small buffer resources.  This is often most noticeable when 3750 QoS is enabled with its default settings.  (NB: I have found, sometimes buffer tuning can mitigate most of the drops.  For example, I had a case of a couple of SAN devices connected to a 3750 having multiple drops a second.  After QoS buffer tuning, I only had a few drops a day.)

As to whether OP question, can a server NIC cause (contribute to) this, it might if Ethernet flow control is enabled and the server is sending pause frames to the switch.

[azibnaseem · Accepted solution]

Hey [@venkat.moturi]  '

From my real time experience output drops can cause by heavy traffic. Long time ago I have faced similar problem & i have changed port & also Ethernet Cable but didn't solved the problem. Then I have reduce some traffic from that interface.

Regards'