Re: SNMPD - MRTG- PPP stats iregularity
On Sat, 24 Oct 1998 08:35:46 +0200, Rainer Clasen wrote: >Yes. 2.1 counts bytes in /proc/net/dev. But IIRC there are still some >NIC-drivers which don't supply these numbers. tulip does. I've no clue about >SNMPD, but I suppose you'll have to teach it to use the byte counters, too. Aiyeee. Looks like it is time for me to upgrade and learn how to use ipchains. :/ -- Steve C. Lamb | I'm your priest, I'm your shrink, I'm your ICQ: 5107343 | main connection to the switchboard of souls. ---+-
Re: SNMPD - MRTG- PPP stats iregularity
Hi! Steve Lamb ([EMAIL PROTECTED]): > On Fri, 23 Oct 1998 19:48:19 -0400, Michael Stone wrote: > > >Linux doesn't ordinarily have any facility for saying _how much_ data > >goes through an if. The snmpd's base their throughput figures on > >_number of packets_, which isn't exactly the same thing. IIRC, you can > >run ip accounting and hack the snmpd's to get their numbers via that > >mechanism. > > Well, that blows. Packets don't mean diddly compared to bytes when it > comes to bandwidth utilization. Do you know if that will be fixed any time > soon? 2.1.* maybe? Yes. 2.1 counts bytes in /proc/net/dev. But IIRC there are still some NIC-drivers which don't supply these numbers. tulip does. I've no clue about SNMPD, but I suppose you'll have to teach it to use the byte counters, too. Rainer -- KeyID=58341901 fingerprint=A5 57 04 B3 69 88 A1 FB 78 1D B5 64 E0 BF 72 EB
Re: SNMPD - MRTG- PPP stats iregularity
On Fri, 23 Oct 1998, Steve Lamb wrote: > On Fri, 23 Oct 1998 19:48:19 -0400, Michael Stone wrote: > > >Linux doesn't ordinarily have any facility for saying _how much_ data > >goes through an if. The snmpd's base their throughput figures on > >_number of packets_, which isn't exactly the same thing. IIRC, you can > >run ip accounting and hack the snmpd's to get their numbers via that > >mechanism. > > Well, that blows. Packets don't mean diddly compared to bytes when it > comes to bandwidth utilization. Do you know if that will be fixed any time > soon? 2.1.* maybe? I use the IP accounting rules in the 2.0.x kernels, works very well with MRTG for real bandwidth stats. Jason
Re: SNMPD - MRTG- PPP stats iregularity
On Fri, 23 Oct 1998 19:48:19 -0400, Michael Stone wrote: >Linux doesn't ordinarily have any facility for saying _how much_ data >goes through an if. The snmpd's base their throughput figures on >_number of packets_, which isn't exactly the same thing. IIRC, you can >run ip accounting and hack the snmpd's to get their numbers via that >mechanism. Well, that blows. Packets don't mean diddly compared to bytes when it comes to bandwidth utilization. Do you know if that will be fixed any time soon? 2.1.* maybe? -- Steve C. Lamb | I'm your priest, I'm your shrink, I'm your ICQ: 5107343 | main connection to the switchboard of souls. ---+-
Re: SNMPD - MRTG- PPP stats iregularity
Quoting Steve Lamb ([EMAIL PROTECTED]): > I've got SNMPD installed on my system so I can monitor, for the moment, > bandwidth usage on eth0 and ppp0. The problem I'm having is that the > incoming and outgoing usages do *NOT* show any divergance. If you want to > see what I mean, take a look at: > http://teleute.ml.org/mrtg/ Linux doesn't ordinarily have any facility for saying _how much_ data goes through an if. The snmpd's base their throughput figures on _number of packets_, which isn't exactly the same thing. IIRC, you can run ip accounting and hack the snmpd's to get their numbers via that mechanism. Mike Stone