04-12-2013 03:19 PM - edited 03-04-2019 07:35 PM
Hi experts,
Here is the config on my router and I am trying to understand the drops for cs6
policy-map llq-policy
class voice-all-out
priority 3000
police 3000000 conform-action transmit exceed-action drop
class class-default
bandwidth remaining percent 90
random-detect dscp-based
random-detect dscp 0 32 60
!
policy-map shaper-out-30m
class class-default
shape average 30000000
service-policy llq-policy
!
interface GigabitEthernet0/0
service-policy output shaper-out-30m
Here is policy-map statistics on the interface Gi0/0. You can go straight to bottom part at the class-default since that's where my question is
#show policy-map int g0/0 output
GigabitEthernet0/0
Service-policy output: shaper-out-30m
Class-map: class-default (match-any)
322074 packets, 349183843 bytes
30 second offered rate 27492000 bps, drop rate 0 bps
Match: any
Queueing
queue limit 64 packets
(queue depth/total drops/no-buffer drops) 19/0/0
(pkts output/bytes output) 322369/349567120
shape (average) cir 30000000, bc 120000, be 120000
target shape rate 30000000
Service-policy : llq-policy
queue stats for all priority classes:
queue limit 64 packets
(queue depth/total drops/no-buffer drops) 0/0/0
(pkts output/bytes output) 8247/1750735
Class-map: voice-all-out (match-any)
8247 packets, 1750735 bytes
30 second offered rate 82000 bps, drop rate 0 bps
Match: ip dscp ef (46)
8247 packets, 1750735 bytes
30 second rate 82000 bps
Priority: 3000 kbps, burst bytes 75000, b/w exceed drops: 0
police:
cir 3000000 bps, bc 93750 bytes
conformed 8247 packets, 1750735 bytes; actions:
transmit
exceeded 0 packets, 0 bytes; actions:
drop
conformed 82000 bps, exceed 0 bps
Class-map: class-default (match-any)
313827 packets, 347433108 bytes
30 second offered rate 27409000 bps, drop rate 0 bps
Match: any
Queueing
queue limit 64 packets
(queue depth/total drops/no-buffer drops) 18/0/0
(pkts output/bytes output) 314124/347819413
bandwidth remaining 90% (24300 kbps)
Exp-weight-constant: 9 (1/512)
Mean queue depth: 17 packets
dscp Transmitted Random drop Tail drop Minimum Maximum Mark
pkts/bytes pkts/bytes pkts/bytes thresh thresh prob
default 314957/347941365 0/0 0/0 32 60 1/10
cs6 228/21186 0/0 0/0 32 40 1/10
My problem is that I once saw the drops for cs6 in the class-default. I cleared the counter but I am sure it will drop again. Based on my understanding it is likely to be routing protocol packets. My question is that, where is the "maximum threshold 40" coming from for cs6? Is it default settings for Random Early Drop? Is there a command I can show the Min and Max thresholds for each DSCP value?
I also only see "default" and cs6 traffic here but not for other DSCP values. Is it because that the router only sees these traffic?
I am wondering if I should simply increase the Min and Max for CS6 or I should create a class for CS6 traffic. Any advice?
Thanks,
04-12-2013 04:27 PM
Hello
you can use these commands to see dscp wred config
show queueing random-detect
show queueing interface xxx
show interfaces
show queue interface-type xx
I cannot see cs6 in the default which is set dscp 0 min threshold to 32 and max to 60
What I don't understand at this time is the voice is llq at 3k and and policed at 3mb on a 30mb cir
And you say you see drops on the default class which looks like its currently set to 27mb but I could be wrong depending on the clock rate of your physical link which is not shown
Res
Paul
Sent from Cisco Technical Support iPad App
04-12-2013 07:21 PM
Thanks Paul for your reply. The commands you provided either gave me no output or told me "the command is depracated."...
r01#show queueing random-detect
Current random-detect configuration:
r01#
r01#show queueing interface g0/0
show queueing command is depracated. Please useshow policy-map interface
r01#show queue g0/0
show queue command is depracated. Please useshow policy-map interface
The physical link is 100mbps. However the circuit provided is only 30mbps service... Thanks
04-12-2013 06:13 PM
Disclaimer
The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.
Liability Disclaimer
In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.
Posting
Based on my understanding it is likely to be routing protocol packets. My question is that, where is the "maximum threshold 40" coming from for cs6? Is it default settings for Random Early Drop? Is there a command I can show the Min and Max thresholds for each DSCP value?I also only see "default" and cs6 traffic here but not for other DSCP values. Is it because that the router only sees these traffic?
I am wondering if I should simply increase the Min and Max for CS6 or I should create a class for CS6 traffic. Any advice?
Likely "maximum threshold 40" for CS6 is the default. Recall it's the default maximum on most REDs. (Also recall, min usually defaults to an increase of two packets per CS. I.e. CS7 might be 34 40 while CS5 might be 30 40, etc.)
Hmm, surprised that show policy map didn't show all the RED stats (usually does on earlier platforms/IOSs). Try just a show policy without reference to the interface.
As you don't want to delay the CS6 traffic, and it's likely routing protocol packets, either create a different class for it or enable FQ in class-default. If you do the latter, also increase its min/max thresholds (as they are global to the class).
04-12-2013 07:27 PM
r01#show policy-map llq-policy
Policy Map llq-policy
Class voice-all-out
priority 3000 (kbps)
police cir 3000000 bc 93750
conform-action transmit
exceed-action drop
Class class-default
bandwidth remaining 90 (%)
packet-based wred, exponential weight 9
dscp min-threshold max-threshold mark-probablity
----------------------------------------------------------
default (0) 32 60 1/10
Still doesn't show me thresholds for the other DSCP values...
I think just to be safe I will create another class and put CS6 and 7 traffic in it and use "bandwidth remaining 10%" to gurantee the bandwidth. One more question, in the config there is also "police 3000000 conform-action transmit exceed-action drop". Do I need it? If I didn't remember wrong, if I use "priority" command it will have implied policing it, is it not right?
Just being curious, to enable FQ for the class-default, I just need to put in "fair-queue" right? What commands do I use to increase the global thresholds...? Is it RED thredholds or FQ thredholds?
Thanks,
04-12-2013 07:35 PM
Hi Difan,
DSCP based WRED automatically setup min/max threshold for only dscp 6 traffic as 32/40....as we know this is mostly network control type of traffic.
So, you can manually configure min/max threshold for other dscp class of traffic e.g. dscp 0 - which you just did
Also, we're not seeing other class of traffic, if you're sending say dscp 3, then it will automatically show up from the show policy-map for the interface
04-12-2013 07:58 PM
Thanks Olayemi. So what is the default min/max threshold for other traffic...? I mean I have RED enabled globally for class-default so there must be a default value for other traffic correct? Thanks!
04-12-2013 08:03 PM
The default are shown below:
dscp Transmitted Random drop Tail drop Minimum Maximum Mark
pkts/bytes pkts/bytes pkts/bytes thresh thresh prob
default 152/10292 0/0 0/0 20 40 1/10
cs1 412/619648 0/0 0/0 22 40 1/10
cs2 244/366976 0/0 0/0 24 40 1/10
cs3 1161/1746144 0/0 0/0 26 40 1/10
cs4 188/282752 0/0 0/0 28 40 1/10
cs5 266/400064 0/0 0/0 30 40 1/10
cs6 142/11972 0/0 0/0 32 40 1/10
cs7 1222/1837888 0/0 0/0 34 40 1/10
Regards
Yemi
Please rate all helpful posts and close solved questions
04-12-2013 08:53 PM
Thank you. So what command did you run to get this output? or do you have to have traffic for all the classes to see them?
04-13-2013 04:50 AM
Disclaimer
The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.
Liability Disclaimer
In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.
Posting
One more question, in the config there is also "police 3000000 conform-action transmit exceed-action drop". Do I need it? If I didn't remember wrong, if I use "priority" command it will have implied policing it, is it not right?
LLQ does police, but only when there's congestion. If there's no congestion, your LLQ can consume more bandwidth. So, whether you need the policer depends on whether you want to enforce the LLQ rate all the time.
Just being curious, to enable FQ for the class-default, I just need to put in "fair-queue" right? What commands do I use to increase the global thresholds...? Is it RED thredholds or FQ thredholds?
Yup, just use fair-queue in class-default (likely on a 3945 FQ is supported in any non-LLQ class).
I was refering to the RED thresholds for CS6, which can be overridden in the manner that default is being configured in your policy.
04-12-2013 08:59 PM
Yes, you need to have the traffic and then it will show up from the show policy-map int x/y output command
Sent from Cisco Technical Support iPhone App
04-12-2013 09:05 PM
Thank you. So my last question is that my priority queue has "police" command in it. I believe people put it in to prevent the queue starves others. However I believe I heard from somewhere that the priority queue itself has implied policing in it. Is it true? Is my "police" command in the queue unnecessary?
04-12-2013 09:17 PM
Priority is just a fancy way of creating a special queue; this queue get serviced first as a latency queue and it is automatically policed only during congestion.
Priority will not police if there is no congestion; whereas Police always drop pckts that exceeds the given value - whether congestion or not
For more info:
http://www.davidsudjiman.info/2011/02/01/qos-priority-and-policing/
Priority is a conditional policer. For example, if you prioritize 48Kbps icmp traffic and there is no congestion, the traffic can utilize the unused bandwidth.
Policing is an unconditional policer. For example, if you police 48Kbps icmp traffic and there is no congestion, policing will still keep the traffic 48Kbps maximum.
Unless you want to make sure the prioritized traffic is using exact amount of bandwidth without having the option to utilize the unused bandwidth, having policing set with priority will make superfluous configuration task.
Sent from Cisco Technical Support iPhone App
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