07-31-2008 10:35 AM - edited 03-03-2019 10:58 PM
I am seeing the following from multiple routers:
WASE520CE2#sh policy-map int mul 1 ou class class-default
Multilink1
Service-policy output: QOS_OUT
Class-map: class-default (match-any)
135086302 packets, 55933251373 bytes
5 minute offered rate 139000 bps, drop rate 0 bps
Match: any
Queueing
Output Queue: Conversation 29
Bandwidth remaining 40 (%)
(pkts matched/bytes matched) 31418640/34103558220
(depth/total drops/no-buffer drops) 0/14882/0
exponential weight: 9
mean queue depth: 0
dscp Transmitted Random drop Tail drop Minimum Maximum Mark
pkts/bytes pkts/bytes pkts/bytes thresh thresh prob
af11 0/0 0/0 0/0 32 40 1/10
af12 0/0 0/0 0/0 28 40 1/10
af13 0/0 0/0 0/0 24 40 1/10
af21 135071422/55924722613 11505/6885849 3377/1643583 32 40 1/10
af22 0/0 0/0 0/0 28 40 1/10
07-31-2008 10:41 AM
Sorry, pressed enter too soon...
The af21 is part of default class so should I adjust the min./max. threshold or increase the interface level output queue to fix this?
What "total drops" really showing as the circuit utilization was not hight at all?
WASE520CE2#sh int mul1
Multilink1 is up, line protocol is up
Hardware is multilink group interface
Internet address is 147.225.39.178/30
MTU 1500 bytes, BW 3072 Kbit, DLY 100000 usec,
reliability 255/255, txload 30/255, rxload 83/255
Encapsulation PPP, LCP Open, multilink Open
Listen: CDPCP
Open: IPCP, loopback not set
Keepalive set (10 sec)
DTR is pulsed for 2 seconds on reset
Last input 00:00:00, output never, output hang never
Last clearing of "show interface" counters 4w5d
Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 14882
Queueing strategy: Class-based queueing
Output queue: 0/1000/64/14882 (size/max total/threshold/drops)
...
07-31-2008 01:37 PM
Hello Kuangyu
Can you please paste the output of "sh queuing interface intfname"
Regards
07-31-2008 01:46 PM
Here is the output... Thanks...
WASE520CE2#sh queueing interface multilink 1
Interface Multilink1 queueing strategy: fair
Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 14945
Queueing strategy: Class-based queueing
Output queue: 0/1000/64/14945 (size/max total/threshold/drops)
Conversations 0/4/16 (active/max active/max total)
Reserved Conversations 5/5 (allocated/max allocated)
Available Bandwidth 2304 kilobits/sec
WASE520CE2#
07-31-2008 07:15 PM
Here is my opinion, hoping to hear from other NetPros as well.
135086302 packets are processed through that interface. A total of 31418640 of these packets are held by software class-default queue because of congestion ocurred time to time. And only a total of 14945 packets are dropped by software queue due to overhad of class-default traffic request in time of congestion. To me, dropping 14945 packets out of 135086302 packets is pretty normal for a class-default traffic which does not contain delay or packet loss sensitive streams.
But the total processed packets counter may be running for long time, and drops might have hit 14945 in recent days or hours. To find it out, I suggest droppping load interval in interface to 30 seconds (You can change back to 5 mins after you are done), clear all statistics related to packets and queuing, then monitor statistics and see if dropping occurs reasonably.
08-01-2008 06:19 AM
Here is the output from this morning:
WASE520CE2>sh queueing interface multilink 1
Interface Multilink1 queueing strategy: fair
Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 15380
Queueing strategy: Class-based queueing
Output queue: 0/1000/64/15380 (size/max total/threshold/drops)
Conversations 0/4/16 (active/max active/max total)
Reserved Conversations 5/5 (allocated/max allocated)
Available Bandwidth 2304 kilobits/sec
The total drops are increasing ands there are 11832 Random drop and 3548 Tail drop pkts.
How should I tune this so there won't be any drops as the circuit utilization is really less than 10% over the night and no congestion.
08-01-2008 05:28 PM
Typically, the type of output drops you're seeing is due to transient congestion. A lot can happen within a few seconds that doesn't appear on a longer term average usage rate.
Since you're using FQ within class-default, you might first try deactivation of WRED. Congestion will then tail-drop per flow.
08-07-2008 07:52 AM
I have disabled the WRED but still seeing drops and not sure if it was from tail-drop as the "show policy-map" output is not classifying the traffic anymore.
Would it helps if I increase the max Output queue?
08-07-2008 10:25 AM
"Would it helps if I increase the max Output queue?"
It may, but avoiding any and all drops can be counter productive. It's the most common way for TCP flows to detect congestion. However, with FQ, at least other flows aren't usually too adversely impacted if you make the total queue too large.
If you do want to increase it, set the maximum queue size within class-default so it's 50% to 100% larger than the default size. This should indirectly increase the individual flow queues.
08-07-2008 10:55 AM
Thanks for the reply.
Is it "Max Threshold for Tail Drop" with "queue-limit" setting you mentioned for increasing the maximum queue size?
Not really see any other command to increase the queue size for class-default:
WASE520CE1(config-pmap-c)#?
QoS policy-map class configuration commands:
bandwidth Bandwidth
compression Activate Compression
drop Drop all packets
estimate estimate resources required for this class
exit Exit from QoS class action configuration mode
fair-queue Enable Flow-based Fair Queuing in this Class
netflow-sampler NetFlow action
no Negate or set default values of a command
police Police
priority Strict Scheduling Priority for this Class
queue-limit Queue Max Threshold for Tail Drop
random-detect Enable Random Early Detection as drop policy
service-policy Configure Flow Next
set Set QoS values
shape Traffic Shaping
08-07-2008 02:59 PM
Should be the "queue-limit" command.
If you do a "show policy-map interface", hopefully you'll see what the default queue limit is (likely 64). If so, try 96 or 128. Show policy-map interface then should show the new setting.
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide