Monitoring System Load (5min avg) on Linux

Monitoring System Load (5min avg) on Linux

Post by Peter Gabrie » Fri, 04 Jul 2003 20:57:35



Hi NG,

the old Problem of logging System Load on Linux.
My Problem is not to gather the data. I have a display Problem.

So:

I get the data multiplied by 100.
My System usually has a low System Load, so i will values like 13, 5, 34, and so on, showing me, that the load is like 0.13, 0.05, etc.

My Problem: I cannot manage to display the values on a graph with the right scaling.

A solution i thought is using YTickScale 0.01, but it doesn't work. Are the values to low for mrtg?

YTickScale 0.1 is working properly, but the values are then multiplied by 10...

Below is my ConfigurationSection for the SystemLoad

Thx,
Peter

-------------------------------------------------

Target[load]: `/root/mrtg/load5.pl`
Title[load]: Auslastung
PageTop[load]: <H1>Auslastung</H1>
MaxBytes[load]: 10000
YLegend[load]: Load
XSize[load]: 350
YSize[load]: 200
YTicsFactor[load]: 0.01
ShortLegend[load]: L
Options[load]: growright, gauge, noo, unknaszero, nopercent, transparent, noinfo

 
 
 

Monitoring System Load (5min avg) on Linux

Post by Andrei Ivan » Sat, 05 Jul 2003 04:08:45



> the old Problem of logging System Load on Linux.
> My Problem is not to gather the data. I have a display Problem.

I'm gathering data differently. Even on the localhost
I have UCD-SNMP daemon running, and so I'm using the following
OID to retrieve 5min load average:

  enterprises.ucdavis.laTable.laEntry.laLoadInt.2
    = .1.3.6.1.4.1.2021.10.1.5.2

Load is reported in % (integers), so if uptime/w reports 0.51,
then snmpget will bring you 51 instead.


  Options[linux-la]: growright,noinfo,nopercent,absolute,gauge,integer
  YLegend[linux-la]: load average, %
  ShortLegend[linux-la]: %

Works in both 2.8.12 and 2.9.6 (MRTG versions).

--
andrei

 
 
 

Monitoring System Load (5min avg) on Linux

Post by Dave Shiel » Sat, 05 Jul 2003 17:36:33



> I'm using the following
> OID to retrieve 5min load average:

>   enterprises.ucdavis.laTable.laEntry.laLoadInt.2
>     = .1.3.6.1.4.1.2021.10.1.5.2

> Load is reported in % (integers), so if uptime/w reports 0.51,
> then snmpget will bring you 51 instead.

That's not strictly correct.  This value is *NOT* a percentage.
As Peter stated, this is the load average multiplied by 100.
This means that it can legitimately exceed 100, if the load
average goes above 1.0

Dave

 
 
 

Monitoring System Load (5min avg) on Linux

Post by Peter Gabrie » Mon, 07 Jul 2003 17:26:27


Quote:> That's not strictly correct.  This value is *NOT* a percentage.
> As Peter stated, this is the load average multiplied by 100.
> This means that it can legitimately exceed 100, if the load
> average goes above 1.0

Yeah, thanks Dave,

that's correct, I think the definition of the computer Load (not CPU load) is somewhat like the simultaneous running processes...

so my problem is just displaying the values, i can get values like 0 (zero load) or 50 (0.5 computer load) or maybe 300 (computer load of 3). I want them displayed to be scaled divided by 100, so that the original values occur...

anybody another idea?
when it is possible to get the 5min avg load of the computer via snmp, i would appreciate...

peter

 
 
 

1. networking linux to a win95 system I am usring redhat 4.0 on my linux system

Greetings ....

I have a linux box which is connected to the net via ISDN (Motorala
Bitsurfer Pro).  When it dials in, it gets the IP 204.255.98.20. I have
'defaultroute' specified in my /etc/ppp/options  file.

I also have a small ethernet network, which up until now, I have used
for
nothing but virtual-hosting.  The IP range is 207.16.115.64 through
207.16.115.127.  The netmask is 255.255.255.192.  All of my virtual
hosts
work beautifully.

The commands I use in /etc/rc.d/rc.local to initialze the ethernet card
on
the linux box are as follows:

ifconfig eth0 207.16.115.65 netmask 255.255.255.192 broadcast
207.16.115.127
route add -net 207.16.115.64 netmask 255.255.255.192 dev eth0

I now desire to hook a win95 box to the linux box via ethernet.  I have
tried every configuration possible on the win95 box, but to no avail.  I
get no ping no matter what I do.  The settings that I believe SHOULD
work
are as follows:

IP Address 207.16.115.125
Gateway 207.16.116.65 (The Linux Box)
Subnet Mask: 255.255.255.192

Am I overlooking something?  I KNOW the hardware and cabling are good as
all were recently extracted from another small (non-internet) network in
the same room.  I am beginning to think that Win95 has a deficiency in
not
being able to deal with small subnets.  I have connected several win95
boxes with linux boxes in the past, but I always had a full Class-C
before.

I apologize if I have given more detail than necessary, but I am to the
point of pulling my hair out.  Any suggestions would be GREATLY
appreciated!

2. comp.dcom.sys.cisco Frequently Asked Questions (FAQ)

3. System-load-monitoring.......

4. Press Ready

5. Any good suggestion for System Monitoring in WinNT2000 and Linux RedHat 7?

6. How does DNS mechanism work?

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

8. Proxy Server Will Not Work After Critical Updates

9. SETTING UP SYSTEM MONITOR TO MONITOR YOUR MODEM

10. System monitor cannot monitor throughput?

11. System Monitor, how to monitor LAN card?

12. Show interfaces 5min in/out rates How to change?