Need some network monitoring tool

Alex_LiquidHost

Hey guys,

I am in need of a tool to monitor the current usage of the network interfaces on my VPS node, current usage and if possible (not a must) some history. Some kind of slave-master configuration would work as well (master web interface to displey the data of all the monitored vps nodes - slaves). RHEL5 and RHEL6 systems.

Thanks in advance.

Jack

vnstat? for live usage.

Observium to record it?

Randy

cacti?

Jack

@Randy said: cati?

Cake

Alex_LiquidHost

As far as I know observium needs its own php server and mysql DB, would it in somehow interact with solusvm running on the node? :)

jhadley

Cacti would work fine

Randy

@jhadley said: Cacti

there we go :)

rds100

Good old mrtg.

nocom

2¢ for RRDtool

KnightSwarm_Phillip

Observium, collectd, cacti...any of a handful of others :)

plueschopath

munin?

ShardHost

Zabbix :)

andreblue

http://www.wireshark.org is a nice one. You can also write lua scripts for it to do custom things.

corehosting

@ShardHost said: Zabbix :)

+1 :)

Raymii

Just plugging: http://raymii.org/cms/p_Bash_PHP_Server_Status_Monitor now has history and saves network traffic. As you can see here: http://vps11.sparklingclouds.nl/status/stat.php the Vserver nodes at Edis.at (vps12 and vps13) (@William) are on 5 TB of TX and almost 3 TB of RX network traffix.

Zen

TX : root@home:~# /sbin/ifconfig eth0 |grep bytes|cut -d":" -f3|cut -d" " -f1:22307021 root@home:~# RX : root@home:~# /sbin/ifconfig eth0 |grep bytes|cut -d":" -f2|cut -d" " -f1:60018058 root@home:~#

Always fun to DIY.. you can get the ongoing rates by comparing the difference between RX/TX @ xtime and ztime

like this :

58702.5 kb 5 seconds later 58702.9 kb

58702.9 minus 58702.5 = 0.4

0.4 / 5 (seconds) = 0.08

0.08 kbps or 81.92bps

Zen

Quick POC without real time computation :

Zen

Real time calculation (5sec/update) based on above math... :

craigb

@alex_liquidhost collectd for stats collection, transmission and threshold monitoring. It's light, coded with security in mind, mature and offers a ton of plugins including libvirt and openvz. Means you can run a single collectd instance on a node and also get stats per container (in addition to the node itself).

Collectd will save the stats in rrd format so there's a bunch of tools that can read that and present graphs, including mrtg and visage (web 2.0 goodness). See here for more: http://collectd.org/related.shtml

Zen

spamming this thread heh

Maybe I'll hook this up to RRDtool, change a few things around, and you've got a working container monitoring tool which will alert you when a VM break a predefined limit ;)

apollo15

@Zen good work!