MRTG and linux load average.

MRTG and linux load average.

Post by Gregers Peterse » Tue, 24 Jun 2003 22:38:43



Hello

I have setup mrtg to monitor memory, swap and disk usage and the traffic on
eth0 etc.
However I'm now stuck with the load average (CPU load and number of
processes).
I have tried with:

As seen on my linux user groups homepage, but it does not seem to work.

If i try "snmptranslate -On -IR" on the OID, it says that linuxLoadAvgOne
does not exist?

If anyone has any idea of what I should try, it will be much appreciated.

My OS is Red Hat linux 7.3.

--
Kindest regards

Gregers Petersen

 
 
 

MRTG and linux load average.

Post by Niels Bagges » Wed, 25 Jun 2003 01:45:34



> However I'm now stuck with the load average (CPU load and number of
> processes).
> I have tried with:


There was never anything calles linuxLoadAvgOne. CMU-SNMP for Linux had
linuxLoadAvg1, but I am quite sure that your RedHat system comes with
UCD-SNMP (or maybe even NetSNMP) and then you should look at laLoadInt.2
which has the five minute load average multiplied by 100. You are also
missing a .0 at the end of hrSystemProcesses, so a useful target could
be


You could also take a trip around
http://net-snmp.sourceforge.net/tutorial-5/mrtg/index.html
for some additional hints about {UCD,Net}-SNMP and MRTG

/Niels

--

The purpose of computing is insight, not numbers  --  R W Hamming

 
 
 

MRTG and linux load average.

Post by Gregers Peterse » Wed, 25 Jun 2003 01:54:12




Quote:> There was never anything calles linuxLoadAvgOne. CMU-SNMP for Linux had
> linuxLoadAvg1, but I am quite sure that your RedHat system comes with
> UCD-SNMP (or maybe even NetSNMP) and then you should look at laLoadInt.2
> which has the five minute load average multiplied by 100. You are also
> missing a .0 at the end of hrSystemProcesses, so a useful target could
> be

Thanks, I can only wonder why SSLUG's mrtg.cfg says hrSystemProcesses
without the zero.

Happy Skt. Hans Aften to you Niels.

--
Regards

Gregers Petersen, Stud. Polyt.

 
 
 

1. MRTG not saving Linux cpu load if load > 100

Hi,
I have:
redhat-release-7.1-1
mrtg-2.9.17-1.rh7
ucd-snmp-4.2-12

and this mrtg.cfg:
WorkDir: /var/www/html/status/mrtg
# load

MaxBytes[load]: 100
Title[load]: Linux load
PageTop[load]: <H1>Linux CPU Load %</H1>
ShortLegend[load]: #
YLegend[load]: Load
Legend1[load]: Avg 1min CPU load
Legend2[load]: Avg 5min CPU load
LegendI[load]: Load 1min
LegendO[load]: Load 5min
Options[load]: growright, gauge

The problem: is that mrtg is not saving in the .log file the history
of the linux load.
The file always always apears cero after the first line.
This happens always that the load of linux is greater than 100 (1.0)

I use this OID:
snmpget localhost public laLoadInt.1
snmpget localhost public laLoadInt.2

The .log file looks like this:
$ head  /var/www/html/status/mrtg/load.log
1000847402 143 140
1000847402 0 0 0 0
1000847101 0 0 0 0
1000847100 0 0 0 0
1000846800 0 0 0 0
1000846500 0 0 0 0
1000846200 0 0 0 0

If I run mrtg every 5minutes, then all the lines get ceros
and the first line get two number similar like this:
1000847702 139 136
1000847702 0 0 0 0
1000847402 0 0 0 0
1000847400 0 0 0 0
1000847100 0 0 0 0
1000846800 0 0 0 0
1000846500 0 0 0 0
1000846200 0 0 0 0

Anyone having this problems?

Thanks
Oliver

2. Beginner Question - error LNK2001: unresolved external symbol __imp__InitCommonControls@0

3. Using MRTG to monitor server load average

4. SIGSEGV error when trying to load file contents into an array

5. CPU Load Linux with MRTG

6. Video won't start!

7. MRTG averages over a time period - mean vs sampling

8. graphics problems

9. MRTG finding a true average util based on operating hours

10. MRTG: Changing Graph Averages

11. MRTG Average Traffic Statistics

12. MRTG problem: Blue lines average 5 samples, green is per sample

13. Computing monthly averages using mrtg logs?