What is the best way to monitor the total incoming / outcoming network traffic of CentOS server. I think that the solution is to monitor the network interfaces and to send SNMP packets to remote server. But is it possible?
regards
On Mon, Dec 21, 2009 at 9:08 AM, sadas sadas mailrc@abv.bg wrote:
What is the best way to monitor the total incoming / outcoming network traffic of CentOS server. I think that the solution is to monitor the network interfaces and to send SNMP packets to remote server. But is it possible?
Absolutely. Check out Cacti http://cacti.net/
There are lots of alternatives to Cacti too (MRTG, even doing graphing in your system monitoring tool like Nagios.) You might also find it easier to monitor the switch instead of the servers if that's possible with your gear. This way, you configure the switch once and can monitor all the servers on it instead of having to configure each server individually to allow snmp access.
Jake wrote:
On Mon, Dec 21, 2009 at 9:08 AM, sadas sadas <mailrc@abv.bg mailto:mailrc@abv.bg> wrote:
What is the best way to monitor the total incoming / outcoming network traffic of CentOS server. I think that the solution is to monitor the network interfaces and to send SNMP packets to remote server. But is it possible?Absolutely. Check out Cacti http://cacti.net/
There are lots of alternatives to Cacti too (MRTG, even doing graphing in your system monitoring tool like Nagios.) You might also find it easier to monitor the switch instead of the servers if that's possible with your gear. This way, you configure the switch once and can monitor all the servers on it instead of having to configure each server individually to allow snmp access.
OpenNMS is good too, if you have enough devices to be worth the extra complexity. Http://www.opennms.org.
On 12/21/2009 9:08 AM, sadas sadas wrote:
What is the best way to monitor the total incoming / outcoming network traffic of CentOS server. I think that the solution is to monitor the network interfaces and to send SNMP packets to remote server. But is it possible?
MRTG is the simplest to setup, but it only does graphs. It's especially easy if you're trying to monitor the local host. You'll need to also install the net-snmp and possibly net-snmp-utils packages.
Network monitoring solutions also do graphs (Cacti, Nagios, OpenNMS).
You can also try ntop. It produces pretty graphs and also segregates network traffic by type/port.
In the past I've used Nagios & NTop. Unfortunately, NTop was a bit of a CPU hog and I had stability issues with it that I never tracked down. So at the moment, we're mostly relying on MRTG to see traffic.
You can also (ab)use MRTG to graph things like CPU usage & CPU temperature, disk utilization, or anything else that you can query via a remote shell command or SNMP query.
On Mon, Dec 21, 2009 at 11:05 AM, Thomas Harold thomas-lists@nybeta.comwrote:
In the past I've used Nagios & NTop. Unfortunately, NTop was a bit of a CPU hog and I had stability issues with it that I never tracked down. So at the moment, we're mostly relying on MRTG to see traffic.
We had the same problems with Ntop (we monitor netflows, not local interfaces). There was a thread about a good netflow tool a week or two ago where someone suggested NfSen + nfdump. I set it up Thursday (took 30 mins or so) and it totally rocks. We've gone to production with it to monitor flows from around 50 devices. I highly recommend you take a look if you have serious needs to see what's really on the wire.
On Mon, Dec 21, 2009 at 12:27 PM, Jake jakepaulus@gmail.com wrote:
We had the same problems with Ntop (we monitor netflows, not local interfaces). There was a thread about a good netflow tool a week or two ago where someone suggested NfSen + nfdump. I set it up Thursday (took 30 mins or so) and it totally rocks. We've gone to production with it to monitor flows from around 50 devices. I highly recommend you take a look if you have serious needs to see what's really on the wire.
I'm going to kick this old thread because I'm just getting around to nfdump and nfsen myself, and am having trouble building nfsen. I do not see a mailing list for it so I'll try here first.
I am using softflowd to pump netflow information from various linux boxes into a central collector. The collector is running nfcapd to capture everything. And I also want to run nfsen on this box to be able to view the pretty graphs on the web :-) I have verified with nfdump built on this box that the collector seems to be working fine - so far so good.
Then when trying to build nfsen it complained about nfprofile not being there - so some googling later and I found that I have to go back and rebuild nfdump to include it.
So I do : ./configure --enable-nfprofile which dies thus : [...] checking for stdint.h... yes checking for unistd.h... yes checking for rrd_update in -lrrd... no configure: error: Can not link librrd. Please specify --with-rrdpath=.. configure failed!
But I check and I see I have librrd installed.
[root@rcadmin:~/nf/nfdump-1.6.1]# rpm -qa | grep -i rrd perl-rrdtool-1.3.7-1.el5.rf rrdtool-1.3.7-1.el5.rf [root@rcadmin:~/nf/nfdump-1.6.1]# ls -al /usr/lib/librrd* -rw-r--r-- 1 root root 321930 May 23 2009 /usr/lib/librrd.a lrwxrwxrwx 1 root root 15 Jun 3 2009 /usr/lib/librrd.so.4 -> librrd.so.4.0.6 -rwxr-xr-x 1 root root 493565 May 23 2009 /usr/lib/librrd.so.4.0.6 -rw-r--r-- 1 root root 319096 May 23 2009 /usr/lib/librrd_th.a lrwxrwxrwx 1 root root 18 Jun 3 2009 /usr/lib/librrd_th.so.4 -> librrd_th.so.4.0.6 -rwxr-xr-x 1 root root 488881 May 23 2009 /usr/lib/librrd_th.so.4.0.6
And I do not seem to be able to tell configure how to find it. These all fail : ./configure --enable-nfprofile --with-rrdpath=/usr/lib ./configure --enable-nfprofile --with-rrdpath=/usr ./configure --enable-nfprofile --with-rrdpath=.. ./configure --enable-nfprofile --with-rrdpath=/usr/lib/librrd.so.4 ./configure --enable-nfprofile --with-rrdpath=/usr/lib/librrd.a
What am I doing wrong?
Whoopsie
[root@rcadmin:~/nf/nfdump-1.6.1]# cat /etc/redhat-release CentOS release 5.2 (Final)
On 21/12/2009 16:05, Thomas Harold wrote:
You can also (ab)use MRTG to graph things like CPU usage& CPU temperature, disk utilization, or anything else that you can query via a remote shell command or SNMP query.
Hi
In this case why not use Ganglia. Look how MediaWiki uses Ganglia with Nagios, and other tools: http://ganglia.wikimedia.org/
Regards
mg.
Whatever you decide to go with, if you are graphing a gigabit interface, make sure you use 64-bit counters. The standard 32-bit counters overflow just past 100Mbit/sec and will give you innacurate readings.
On 12/21/09, Marcelo M. Garcia marcelo.maia.garcia@googlemail.com wrote:
On 21/12/2009 16:05, Thomas Harold wrote:
You can also (ab)use MRTG to graph things like CPU usage& CPU temperature, disk utilization, or anything else that you can query via a remote shell command or SNMP query.
Hi
In this case why not use Ganglia. Look how MediaWiki uses Ganglia with Nagios, and other tools: http://ganglia.wikimedia.org/
Regards
mg. _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
-
Alan McKay -
Jake -
Les Mikesell -
Marcelo M. Garcia -
sadas sadas -
Thomas Harold