Hi all...
I am using net-snmp 5.7.2
CentOS 6.
In my system there are thousands of PPPoE users / virtual inteface.
So snmpd service cause too much load in my system.
Can anyone help me with this ? how to resolve it ?
Which MIB is affected with this ? so I can remove / change code of those
problem...but many snmp softwares have this
problem.
I have patched "net-snmp-5.2.1-correct_cpu_usage.patch" to
net-snmp-5.2.6.rc2.
It made the CPU usage better,but the value of usage is higher than v5.5
15~25% in the most time.
- Original Message -
From: "Wes Hardaker
>>>>> On Fri, 21 May 2010 18:34:20 +0800, "Brian Lu" said:
BL> How can i get correct cpu usage value from ver:5.2.x ?
BL> Any suggestions will be appreciated!!!
The best thing to do, of course, is upgrade!
Or compare the two releases and find the patch th
Hi All
I found the CPU usage can be shown almost correct on ver:5.5
but the cpu usage value is very high on ver:5.2.x
How can i get correct cpu usage value from ver:5.2.x ?
Any suggestions will be appreciated!!!
Best regards,
Brian Lu
Sorry that I have not given all the details.
The OS is RHEL WS 4, 64-bit.
Thanks
Reddy
Dave Shield <[EMAIL PROTECTED]> wrote:
On 13/12/06, Venkata Guddeti wrote:
> I have recently upgraded Net-snmp to 5.4.
> I was getting values for all the columns in the hrStorageTable before with
>
On 13/12/06, Venkata Guddeti <[EMAIL PROTECTED]> wrote:
> I have recently upgraded Net-snmp to 5.4.
> I was getting values for all the columns in the hrStorageTable before with
> 5.3.1 (not sure of the correctness). After upgrade, I get some columns in a
> row missed. For example, values for hrSt
Dear Dave, Thomas and Other coders,
I have recently upgraded Net-snmp to 5.4.
I was getting values for all the columns in the hrStorageTable before with
5.3.1 (not sure of the correctness). After upgrade, I get some columns in a row
missed. For example, values for hrStorageUsed.8 hrStorageUse
Venkata Guddeti wrote:
> I have tried with the options given below.
> But make install still tries to write files outside in the installation dir
> I provided.
> The command I have used for configure is as follows:
> ./configure LDFLAGS=-L/usr/lib64 [EMAIL PROTECTED] --with-sys-location=Lab
>
Thank you very much
I have tried with the options given below.
But make install still tries to write files outside in the installation dir I
provided.
The command I have used for configure is as follows:
./configure LDFLAGS=-L/usr/lib64 [EMAIL PROTECTED] --with-sys-location=Lab
--with-logfile
Dear Dave and All
Thank you very much for the reply. I have tried installing on test machine, I
have got some values, not sure of the correctness.
I have tried to compile the net-snmp 5.4 code on a machine where I do not root
access. This needs root previleges as compilation tried to create d
Dave Shield wrote:
> On 05/12/06, Venkata Guddeti <[EMAIL PROTECTED]> wrote:
>> Is there an option to choose for those files to go into a different dir (of
>> my choice)?
>
> Not sure, offhand.
> But disabling the perl stuff might help.
> Try
> configure --with-out-perl-modules --disable-e
On 05/12/06, Venkata Guddeti <[EMAIL PROTECTED]> wrote:
> I have tried to compile the net-snmp 5.4 code on a machine where I do not
> root access. This needs root previleges as compilation tried to create
> dir/files under the following dir.
> /usr/lib64/perl5/site_perl/5.8.5/x86_64-linux-thread-m
On 01/12/06, Venkata Guddeti <[EMAIL PROTECTED]> wrote:
> It seems that the hrProcessorLoad does not give anything.
> Has anyone successfully used this variable from the HOST-RESOURCES-MIB MIB.
Please try the new 5.4 release. The agent finally implements
this object in a sensible manner (for the
Dear All
It seems that the hrProcessorLoad does not give anything.
Has anyone successfully used this variable from the HOST-RESOURCES-MIB MIB.
I though that I can use ssCpuIdle, but it is already marked as 'deprecated'.
Your reply is greatly appreciated.
Thank you very much,
Reddy
--
Justin Joseph wrote:
I am finding the percentage cpu usage of a process from SNMP
MIB using hrSWRunPerfCPU oid. In case of single processor systems it is
working fine. But it is giving incorrect values in case of systems
having multiple CPUs. In the description of the oid it is
I am finding the percentage cpu usage of a process from SNMP MIB using hrSWRunPerfCPU oid. In case of single processor systems it is working fine. But it is giving incorrect values in case of systems having multiple CPUs. In the description of the oid it is mentioned as "
Note that
nd that just repeats over and over again like I would expect a daemon to.
Anyone got any idea on this?
Jayson Robinson
From: Thomas Anders <[EMAIL PROTECTED]>
To: [email protected]
Subject: Re: snmptrapd cpu usage on solaris when disconnected from pty.
Date: Tue, 11 Apr 2
All I've got a wierd one.
I'm running snmptrapd 5.3.01 on a solaris 9 box. When I run from the
command line and use the -f option I see less then 1% cpu usage. Whenever I
call the snmptrapd from the startup script (and it's using the same exact
options) cpu goes way up:
On Tue, 2006-02-07 at 12:17 -0800, Tony Cratz wrote:
> Is it possible to
> add the UCD-SNMP-MIB.txt to the default Sun snmpdx server?
Not directly - no.
The easiest approach is probably to run the two agents
at the same time, with the Net-SNMP agent being a subagent
of the Sun one (or vice versa)
Hello,
I'm looking for some help. We are using the Denika software to
monitor and graph out systems instead of MRTG. Is it possible to
add the UCD-SNMP-MIB.txt to the default Sun snmpdx server? Or is
there another MIB which can be added instead?
Please r
Hi Dave,
I have a multiprocessor system and I am trying to calculate cpu usage
of a particular process by the formulae given by you below,
But I am getting more than 100%
Later I saw the properties of that OID (hrSWRunPerfCPU) which is "
The number of centi-seconds of the
Thanks, I will try it out.
-Original Message-
From: Robert Story [mailto:[EMAIL PROTECTED]
Sent: 15 October 2005 01:10
To: [EMAIL PROTECTED]
Cc: [email protected]
Subject: Re: snmpd 100% cpu usage
On Fri, 14 Oct 2005 15:24:31 +0100 David wrote:
DW> 32 bit, ker
On Fri, 14 Oct 2005 15:24:31 +0100 David wrote:
DW> 32 bit, kernel 2.6.9-5.ELsmp.
DW>
DW> I found the offending request (StorageAllocationFailures.1) coming from
DW> third party monitoring software. When I snmpwalk walk hrStorage,
DW> StorageAllocationFailures.1 doesn't exist.
That rings lots
-Original Message-
From: Robert Story [mailto:[EMAIL PROTECTED]
Sent: 13 October 2005 23:32
To: [EMAIL PROTECTED]
Cc: [email protected]
Subject: Re: snmpd 100% cpu usage
On Thu, 13 Oct 2005 17:36:05 +0100 [EMAIL PROTECTED] wrote:
WN> I've run into a situation where I hav
up on another box and get the gdb trace.
-Original Message-
From: Robert Story [mailto:[EMAIL PROTECTED]
Sent: 13 October 2005 23:32
To: [EMAIL PROTECTED]
Cc: [email protected]
Subject: Re: snmpd 100% cpu usage
On Thu, 13 Oct 2005 17:36:05 +0100 [EMAIL PROTECTED] wrot
On Thu, 13 Oct 2005 17:36:05 +0100 [EMAIL PROTECTED] wrote:
WN> I've run into a situation where I have snmpd utilising 99.9% of my CPU
WN> continually and failing to service any snmp requests. Running strace on
WN> the process I get the following:
WN>
WN> open("/etc/mtab", O_RDONLY)
> On Thu, 13 Oct 2005 17:36:05 +0100, [EMAIL PROTECTED] said:
wardd> I've run into a situation where I have snmpd utilising 99.9% of
wardd> my CPU continually and failing to service any snmp requests.
wardd> Running strace on the process I get the following:
wardd> open("/etc/mtab", O_RDONLY)
Hello,
I've run into a situation where I have snmpd utilising 99.9% of my CPU
continually
and failing to service any snmp requests. Running strace on the process
I get the following:
open("/etc/mtab", O_RDONLY) = 15
close(15) = 0
open("/etc/mtab", O_RD
lling frequency accordingly.
Thanks,
Naga
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Dave
Shield
Sent: Tuesday, October 04, 2005 2:10 PM
To: [EMAIL PROTECTED]
Cc: [email protected]
Subject: Re: Regarding percentage CPU Usage by a given pr
On Tue, 2005-10-04 at 09:23 +0530, Naganarasimha wrote:
> How to find current value of percentage of CPU utilized
> by a given process?
> hrSWRunPerfCPU will give the “The number of centi-seconds of the
> total system's CPU resources consumed by the process”
Sample this value twice (separated by
Hi ,
How to find current value of percentage of
CPU utilized by a given process?
Similar to the output got in the Task Manager
in Windows.
hrSWRunPerfCPU
will give the “The number of centi-seconds of the total system's CPU resources
consumed by the process” ( CPU Time in TaskMan
Dave Shield wrote:
> Ummm
> What exactly is wrong with these figures?
Check out the sample graph from the NET-SNMP tutorial:
http://net-snmp.sourceforge.net/tutorial/tutorial-4/mrtg/kontor.cpu.html
Compare that with 'My' user vs. idle:
http://cytrex.cytelcom.com/cgi-bin/14all.cgi?
On Thu, 2005-08-25 at 15:19 -0500, Matthew Boehm wrote:
> I have these setup just like in the tutorial, but I am getting wierd
> results:
>
> UCD-SNMP-MIB::ssCpuRawUser.0 = Counter32: 3597620
> UCD-SNMP-MIB::ssCpuRawNice.0 = Counter32: 839
> UCD-SNMP-MIB::ssCpuRawSystem.0 = Counter32: 19131260
>
Hey guys,
I'm trying to graph my cpu usage according to the following tutorial:
http://net-snmp.sourceforge.net/tutorial/tutorial-4/mrtg/
I have these setup just like in the tutorial, but I am getting wierd
results:
UCD-SNMP-MIB::ssCpuRawUser.0 = Counter32: 3597620
UCD-SNM
34 matches
Mail list logo
