Hi
I have a CMTS Casa Systems C100G and I want to understand the output of this command.
CASA-C100G>show cpuinfo module 0
Module 0:
cpu model XLP432 Rev B2 FPU @1200000kH BogoMIPS 268.13
microsecond timers yes
tlb_entries 544
17:27:51 up 3 days, 4:47, 0 users, load average: 0.24, 0.24, 0.25
33676723 processes: 33539139 sleeping, 137584 running, 0 zombie, 0 stopped
CPU states: 24.4% user, 0.0% system, 0.0% nice, 75.5% idle
Data CPU utilization(Min/Avg/Max): (4.0%/37.8%/95.0%)
Mem: 3123456K total, 1590720K used, 1532736K free, 277248K buffers
What is the use of the processor of the QAM module 0, someone help me to interpret this result
Each Line Card on a Casa CMTS is its very own linux integrated system.
As such this command is showing the equivalent of running the linux (unix) command top on the card in question. google search man top command to understand what all the pieces mean.
but I can tell you that it is pretty much idle (75.5%) and not really doing much.
is there some issue your experiencing that your trying to troubleshoot?
We are having problems with the upload of video by WhatsApp, they do not upload completely
reviewing technical support they tell us that it is because the percentage of utilization of the QAM module is at 98% .. so I want to interpret the command to know what the percentage of utilization is in real time
Data CPU utilization (Min / Avg / Max): (4.0% / 37.8% / 95.0%)
Hope you don't mind, I work for Casa as a Systems Engineer and I reviewed the case you have open with Casa.
Few things to note:
I would reccomend a software upgrade to 7.2.6.x
ddos protection is not turned on, on the chassis (it is not supported on 7.2.4.2, but is available for 7.2.6.x)
it is enabled with the ddos command
It prevents your CMTS from being the target of a ddos attack as well as the source of a ddos attack, I would upgrade just for this feature.
ddos
This command enables or disables mitigation of Distributed Denial of Service (DDoS) breaches such as a DNS amplification attack or traffic oversubscription that can potentially limit service flows. The threshold is five simultaneous attacks with a reaction time of less than a second. When enabled, the feature functions as follows:
• Handles oversubscription messages from QAM and drops packets if necessary.
• Programs the switch with meter entries that qualify on destination IPv4 and IPv6 addresses. The meter is the aggregate of all service flows associated with the destination address.
• Maintains the current state of programmed entries in the switch along with current drop counters. (Drop counters are retrieved every five seconds.) • Ages out existing entries when the drop counter on a corresponding entry no longer increases. (Aging out is after 10 minutes by default.)
This has been in our Security best practices since its inception, and if your experiencing the problems it targets, effectively solves the issue.
Your CMTS is pretty loaded, with many upstreams and downstreams at 80% utilization+, I would recommend adding at least 1 more DS card, and moving some customers over to it, in the process splitting up the larger (customer wise and utilization wise) service groups. Adding an 8x96 should be fine. You can combine any DS with any US on the CASA so should be pretty easy CMTS wise, obviously depending on how your headend is combined that may or may not be easy otherwise.
Consider upgrading from 2x10G SMMs to 8x10G SMMs, the 8x10G SMMs have a much more powerful packet processor.
Many MSOs have a 60,70,80 utilization report. I would consider utilizing something similar. basically you keep track of utilization on all of your ports, when the utilization on a particular port hits 60% you make plans to correct it (node splits, additional capacity channels/cards) when it hits 70 you make sure these plans are started, when it hits 80% it becomes your top priority. Reviewing it bi-weekly or monthly seems to be the common approach. but the idea is to not let any port (US or DS) get to the point where it gets anywhere near 80-90% because that is when customers will start noticing issues.
Adding more DS per service group could also help, the bulk of your downstreams are quite utilized.
To bring down the Qam card utilization specifically you could also mark some of your DS channels as non-primary with the secondary key word
Dear kwesibrunee, can you send me a 7.2.6.x software for Casa 100G, my email = leohilbert@gmail.com
Big thanks and best regards.
Sorry, I don't work directly for Casa anymore and don't have access to older software images any more.
7.2.6.x is quite old, like 2-3 years, why are you looking for that image specifically? Unless you have hardware constraints (SMM 2x10G or DS 8x96 cards) I would highly recommend you use a newer 8.6 or 8.8 release. 8.8.3.4 is one of the latest versions and is pretty well battle tested.
Hi
Please help me understand this:
Module 0 is QAM:
Module 0 QAM_8x96 Running (8 ports, 36 channels/port, 96 shared channels)
Docsis channel: 144
Video channel: 0
Broadcast channel: 0
Major rev 5, Minor rev 15
CASA-C100G#show cpuinfo module 0
Module 0:
cpu model XLP432 Rev B2 FPU @1200000kH BogoMIPS 268.13
microsecond timers yes
tlb_entries 544
11:50:40 up 5 days, 22:56, 0 users, load average: 0.24, 0.24, 0.24
62688399 processes: 62432318 sleeping, 256081 running, 0 zombie, 0 stopped
CPU states: 24.6% user, 0.0% system, 0.0% nice, 75.3% idle
Data CPU utilization(Min/Avg/Max): (4.0%/38.2%/94.0%)
Mem: 3123456K total, 1591040K used, 1532416K free, 277248K buffers
The value of 75.3% idle, is the percentage of the CPU that is idle, it is high value and this is good.
load average: 0.24, 0.24, 0.24, is the Average System Load, calculated in 1, 5, and 15 minute intervals
0.24 is a low average, this is good.
Data CPU utilization(Min/Avg/Max): (4.0%/38.2%/94.0%)
This line shows a maximum of 94.0%, what is the period of time to calculate these results ?, this maximum does not coincide with the previous data that shows low processor load, why is this ?
regards
Data CPU shows Datapath CPU utilization over uptime of the module
you can use the show cpu-history module 0 command to show the last 60 mins
CMTS#show cpu-history module 0
< Module00 CPU Usage History >
(%)
100|------------------------------------------------------------
95|
90|
85|
80|
75|
70|
65|
60|
55|
50|
45|
40|
35|
30|
25|
20|
15|
10|XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
05|
00|
=============================================================
0 10 20 30 40 50 60 (min)
you can use the CASA-ENTITY-EXT-MIB to graph the CPU values over time see attached mib files
the casaModuleCpuMemTable will give you the CPU stats, 1,5,15 min avgs + memory usuage and the casaModuleDataCpuUtilizationTable will give you the min/avg/max since the card was booted.
Hi
This is the result:
CASA-C100G#show cpu-history module 0
< Module00 CPU Usage History >
(%)
100|------------------------------------------------------------
95|
90|
85|
80|
75|
70|
65|
60|
55|
50|
45|
40|
35|
30|XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
25|
20|
15|
10|
05|
00|
=============================================================
0 10 20 30 40 50 60 (min)
with this it can be concluded that the use of the processor is on average 30%
The question is why that maximum of 95% on this line:
CASA-C100G#show cpuinfo module 0
Data CPU utilization(Min/Avg/Max): (4.0%/37.7%/95.0%)
in a short time we will use more docsis channels in QAM_8x96 modules, I want to be sure that this does not generate problems as it has a maximum of 98% but it does not match what it shows:
CASA-C100G#show cpu-history module 0
I am attentive to your comments, thanks for your response and help
regards
show cpu-history module 0 only shows the last 60 mins....
Data CPU min average max shows the data for the amount of time the module has been on, think of it as more long term version of show cpu-history module 0.
if your going to use the history command you will need to monitor it during peak time, to see if your hitting that 95% currently.
The number of channels on a port has less to do with CPU spiking than how full the channels are. So adding more channels is likely to reduce your CPU Usage
Hi
We are having problems sending video by WhatsApp through cable modem, some do not finish sending completely
The same to upload to whatsapp states
testing our internet through a router that does not go through cmts works fine
According to the output from show cpu-history, this problem is not due to high processor usage
How can we solve this problem?
Regards
Hello
this is an example of a configuration file to have 10 Mb Down and 2Mb Up
Please review and suggest that I can modify to test if it improves the uploading of videos by whatsapp
GlobalPrivacyEnable 1;
UsServiceFlow
{
UsServiceFlowRef 1;
QosParamSetType 7;
TrafficPriority 0;
MaxTrafficBurst 20000;
MaxRateSustained 2048000;
}
DsServiceFlow
{
DsServiceFlowRef 2;
QosParamSetType 7;
TrafficPriority 0;
MaxTrafficBurst 8000;
MaxRateSustained 10240000;
}
MaxCPE 4;
SnmpMibObject docsDevNmAccessStatus.1 Integer 4; /* createAndGo */
SnmpMibObject docsDevNmAccessIp.1 IPAddress 10.10.0.1 ;
SnmpMibObject docsDevNmAccessIpMask.1 IPAddress 255.255.255.255 ;
SnmpMibObject docsDevNmAccessControl.1 Integer 2; /* read */
SnmpMibObject docsDevNmAccessInterfaces.1 String "@" ;
SnmpMibObject docsDevNmAccessCommunity.1 String "public" ;
SnmpMibObject docsDevNmAccessStatus.2 Integer 4; /* createAndGo */
SnmpMibObject docsDevNmAccessIp.2 IPAddress 180.80.80.20 ;
SnmpMibObject docsDevNmAccessIpMask.2 IPAddress 255.255.255.255 ;
SnmpMibObject docsDevNmAccessControl.2 Integer 3; /* readWrite */
SnmpMibObject docsDevNmAccessInterfaces.2 String "@" ;
SnmpMibObject docsDevNmAccessCommunity.2 String "public" ;
SnmpMibObject docsDevNmAccessStatus.3 Integer 4; /* createAndGo */
SnmpMibObject docsDevNmAccessIp.3 IPAddress 10.8.0.1 ;
SnmpMibObject docsDevNmAccessIpMask.3 IPAddress 255.255.255.255 ;
SnmpMibObject docsDevNmAccessControl.3 Integer 2; /* read */
SnmpMibObject docsDevNmAccessInterfaces.3 String "@" ;
SnmpMibObject docsDevNmAccessCommunity.3 String "public" ;
/* CmMic f1b3eaa1b54fe85bd59e6f096ba3620b; */
/* CmtsMic fd394e2434c0abae65b0b91b3817c6dc; */
/*EndOfDataMkr*/
}
regards