diagnose output drops

[Marcus Johansson]

Hi,

I have some problems on my newly purchased wan routers Cisco 2921 with 15.1T3

on the wan interface is see "ouput drops" this is gigabit ethernet on both sides , but the ISP delivers 30mbit/s on this line.

Any ideas what could be causing this, and how i can prevent it from happening?

ellslr1#sh int gig 0/2
GigabitEthernet0/2 is up, line protocol is up
  Hardware is CN Gigabit Ethernet, address is 1cdf.0f26.beb2 (bia 1cdf.0f26.beb2)
  Description: *** WAN ***
  Internet address is x.x.x.x/x
  MTU 1500 bytes, BW 1000000 Kbit/sec, DLY 10 usec,
     reliability 255/255, txload 1/255, rxload 1/255
  Encapsulation ARPA, loopback not set
  Keepalive set (10 sec)
  Full Duplex, 1Gbps, media type is RJ45
  output flow-control is unsupported, input flow-control is unsupported
  ARP type: ARPA, ARP Timeout 04:00:00
  Last input 00:00:00, output 00:00:00, output hang never
  Last clearing of "show interface" counters never
  Input queue: 0/75/0/1743 (size/max/drops/flushes); Total output drops: 239812
  Queueing strategy: Class-based queueing
  Output queue: 0/1000/0 (size/max total/drops)
  5 minute input rate 2119000 bits/sec, 1163 packets/sec
  5 minute output rate 3750000 bits/sec, 1243 packets/sec
     1150979897 packets input, 665881762 bytes, 4 no buffer
     Received 111240 broadcasts (0 IP multicasts)
     0 runts, 0 giants, 3 throttles
     0 input errors, 0 CRC, 0 frame, 0 overrun, 0 ignored
     0 watchdog, 111239 multicast, 0 pause input
     1305605171 packets output, 1678127137 bytes, 0 underruns
     0 output errors, 0 collisions, 0 interface resets
     111239 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

[Mahesh Gohil]

HI,

not only output drop is to notice but it seems flushes (Input queue: 0/75/0/1743 (size/max/drops/flushes) is increasing on interface.

Can you try clearing interface counter and see how frequently it is increasing.

If still increasing you can increasing hold-queue to span the queue to cater more packets on interface

hold-queue 4096 in

hold-queue 4096 out

Regards

Mahesh

[Marcus Johansson]

tried to clear interface yesterday, so after about 15h some drops are recorded.

but would it be safe to change the hold queue when this wan link is used to transport SIP/VOIP?

GigabitEthernet0/2 is up, line protocol is up
  Hardware is CN Gigabit Ethernet, address is 1cdf.0f26.beb2 (bia 1cdf.0f26.beb2)
  Description: *** WAN ***
  Internet address is x.x.x.x/x
  MTU 1500 bytes, BW 1000000 Kbit/sec, DLY 10 usec,
     reliability 255/255, txload 1/255, rxload 1/255
  Encapsulation ARPA, loopback not set
  Keepalive set (10 sec)
  Full Duplex, 1Gbps, media type is RJ45
  output flow-control is unsupported, input flow-control is unsupported
  ARP type: ARPA, ARP Timeout 04:00:00
  Last input 00:00:02, output 00:00:00, output hang never
  Last clearing of "show interface" counters 15:14:34
  Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 248
  Queueing strategy: Class-based queueing
  Output queue: 0/1000/0 (size/max total/drops)
  5 minute input rate 907000 bits/sec, 483 packets/sec
  5 minute output rate 1168000 bits/sec, 494 packets/sec
     28931329 packets input, 972665336 bytes, 0 no buffer
     Received 914 broadcasts (0 IP multicasts)
     0 runts, 0 giants, 0 throttles
     0 input errors, 0 CRC, 0 frame, 0 overrun, 0 ignored
     0 watchdog, 914 multicast, 0 pause input
     26349491 packets output, 2403418074 bytes, 0 underruns
     0 output errors, 0 collisions, 0 interface resets
     914 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

[Tharak Abraham]

Marcus,

Increasing the hold-queue is not recommended.

Are the users complaining of decreased performance, jitter, blackouts etc ?

You may have to fine tune your shaper so that delay sensitive traffic is not affected.

Ensure that your applications gets the second best throughput.

You may have to drill down to the policy map and see which class is choked ?

And if its the default-class, i would tell just chuck it.

After all 284 www/yahoo/google/public mail traffic drops in 15 hours is just fine !!

[mlund]

Hi Marcus

Can You share the configuration of the g0/2 interface.

Is there any policy-map applied, if it does, can You do show policy-map int g0/2.

/Mikael

[Mahesh Gohil]

Hi,

If you are running time sensitive application it is not safe to increase the hold-queue.

But the question is  this drop is really concern to you. Do you face any degrade in application due to this. If you want you can fine tune

hold queue keeping application status in  mind.

regards

Mahesh

[gfcisco31]

if you have a policy configured for the interface, here´s a good command to see if you have drops in any of the classes you have.

show policy-map interface "seial0/0" | inc Class|offered|drops

if you have drops under the classes, see if it matches with the output drops, it might me related to queue size and/or wred thresholds configured.

hope that helps