Hi, I've used rrdtool before to create stats on my network interfaces but, as all the machines I generate data for reboot quite regularly (desktops, test machines, broken things, etc), I get alot of spikes on my graphs. All are running Linux, either 2.4.24 or 2.6.
I'm looking for the simplest way to graph network usage on these machines without having to worry about manually correcting these spikes. One way I thought would be simple would be to have some sort of utility/daemon that counts packets on interfaces that can reset the count when queried (ie when the rrd update script is run). This way I could use a GAUGE/ABSOLUTE instead of a COUNTER. I believe I could still get packets/sec and bytes/sec (averaged over 5 minutes, of course) from this, but please correct me if I'm wrong. Up until now, I've grabbed information from the output of ifconfig/netstat -i but this quite obviously can't be reset (and I'd rather avoid resetting these, too). I guess running an SNMP daemon on the host would suffer the same issues as the counts would reset when restarted. Are there any simple ways of doing what I'm trying to? Or should I just use a COUNTER and live with the overflows? Graphing data such as temperature seems to be easy, I just can't figure out this. Thanks, David -- Unsubscribe mailto:[EMAIL PROTECTED] Help mailto:[EMAIL PROTECTED] Archive http://www.ee.ethz.ch/~slist/rrd-users WebAdmin http://www.ee.ethz.ch/~slist/lsg2.cgi
