Thinking of adjusting the output queue on 6509 switchport

alalli · ‎11-25-2013

Good Day All,

I have a 6509 switchport that is taking some output errors.

When I monitor the port, it seems to always say that there are 0 packets in the output queue; but then it will drop packets (80-150) at a time.

The port is configured (factory default) at 40 for the output queue.

I am thinking of using the hold-queue output command on the interface to change the size of the output queue.

I was hoping to hear from anyone who has needed to do this in a production environment.

Thanks very much,

Please find a show interface and show run for the interface below (thanks)

GigabitEthernet2/2/11 is up, line protocol is up (connected)

Hardware is C6k 1000Mb 802.3, address is XXXX.XXXX.XXXX. (bia BBBB.BBBB.BBBB

Description: *** Riverbed Management Port ***

MTU 1500 bytes, BW 1000000 Kbit, DLY 10 usec,

reliability 255/255, txload 5/255, rxload 12/255

Encapsulation ARPA, loopback not set

Keepalive set (10 sec)

Full-duplex, 1000Mb/s, media type is 10/100/1000BaseT

input flow-control is off, output flow-control is on

Clock mode is auto

ARP type: ARPA, ARP Timeout 04:00:00

Last input never, output 22w4d, output hang never

Last clearing of "show interface" counters 2y24w

Input queue: 0/2000/1/0 (size/max/drops/flushes); Total output drops: 1059339

Queueing strategy: fifo

Output queue: 0/40 (size/max)

5 minute input rate 50766000 bits/sec, 8926 packets/sec

5 minute output rate 20512000 bits/sec, 6135 packets/sec

446327969 packets input, 296484722646 bytes, 0 no buffer

Received 357217 broadcasts (0 multicasts)

457 runts, 1 giants, 0 throttles

6660 input errors, 2130 CRC, 0 frame, 1 overrun, 0 ignored

0 watchdog, 0 multicast, 0 pause input

The other end is hard coded for 1Gig / full as well.

interface GigabitEthernet2/2/11

description *** Riverbed***

switchport

switchport access vlan QQQ

switchport mode access

switchport nonegotiate

no logging event link-status

speed 1000

duplex full

no cdp enable

spanning-tree portfast edge

end

acampbell · ‎11-26-2013

Hi,

I would toch nothing until you can baseline what the drops are like.

Look at you show int, the counters have not been cleared for over 2 years

" Last clearing of "show interface" counters 2y24w "

start by clear counters g2/2/11

Then take shapshots "show int g2/2/11" say for a week 4 times a day.

Look at the errors before making any judgement.

Regards,
Alex.
Please rate useful posts.

Regards, Alex. Please rate useful posts.

alalli · ‎11-26-2013

Hi Alex,

Many thanks for the reply.

The interface was cleared yesterday. (shortly after sending in the question)

We have taken 46,894 output drops in the last 26 hours.

I monitored the interface just now between 11:30 and 11:38 and it did not drop any packets until 11:38 when it dropped aproximately 100 packets.

I was attempting to monitor the output queue the whole time and the size of the queue remained at 0 every time i ran the "show int gi2/2/11 |include max command"

The interface seems to be quite happy for awhile , and then it drops 100 or so packets, and then seems quite happy.

My idea is to increase the output queue to about 150 (from 40) to accomidate this burst and see if that resolves the issue.

I will have to keep an eye on the CPU, but I would not think that increasing the size of the output queue on a single interface would spike the CPU by much.

I have never had to alter the output queue before and am indeed not sure that changing it will fix this problem, but I think it is worth a try. I was hoping to hear back from anyone who has needed to alter the output queue size and compare notes.

thanks again

regards

Amanda

dc1006509ve01#show int gi2/2/11

GigabitEthernet2/2/11 is up, line protocol is up (connected)

Hardware is C6k 1000Mb 802.3, address is xxxx.xxxx.xxxx.xxxx (bia yyyy.yyyy.yyyy.yyyy)

Description: *** Riverbed Management Port ***

MTU 1500 bytes, BW 1000000 Kbit, DLY 10 usec,

reliability 255/255, txload 4/255, rxload 9/255

Encapsulation ARPA, loopback not set

Keepalive set (10 sec)

Full-duplex, 1000Mb/s, media type is 10/100/1000BaseT

input flow-control is off, output flow-control is on

Clock mode is auto

ARP type: ARPA, ARP Timeout 04:00:00

Last input never, output 22w5d, output hang never

Last clearing of "show interface" counters 1d02h

Input queue: 0/2000/0/0 (size/max/drops/flushes); Total output drops: 46894

Queueing strategy: fifo

Output queue: 0/40 (size/max)

5 minute input rate 38238000 bits/sec, 6968 packets/sec

5 minute output rate 16370000 bits/sec, 4900 packets/sec

447157584 packets input, 302831873482 bytes, 0 no buffer

Received 213 broadcasts (0 multicasts)

0 runts, 0 giants, 0 throttles

0 input errors, 0 CRC, 0 frame, 0 overrun, 0 ignored

0 watchdog, 0 multicast, 0 pause input

0 input packets with dribble condition detected

341850828 packets output, 178099556505 bytes, 0 underruns

0 output errors, 0 collisions, 0 interface resets

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

dc1006509ve01#

dc1006509ve01#show int gi2/2/11 | include max

Input queue: 0/2000/0/0 (size/max/drops/flushes); Total output drops: 46995

Output queue: 0/40 (size/max)

dc1006509ve01#