Multiple IP MRTG or Similar
lists at visionsix.com
Wed Jul 6 11:35:12 GMT 2005
> Hi my expirience is that bandwithd does a terrible job, it sets your
> interface in promisious mode if I'm not mistaken and keeps it's data in
> ram. use it only for investigation or test environment's not on
> production systems (my advice.)
I would like to share some of our experiences with Bandwidthd. As I mentioned in my previous email, Netflow and Bandwidthd are certainly two vastly different applications. However, we have used and tested Bandwidthd in a variety of small and small-medium size environments. In the end we found no problem running Bandwidthd and in each case it did exactly what we needed it to do.
We have ran Bandwidthd on a FreeBSD firewall/ bridge with over 2000 client machines passing traffic through it daily for six months. The clients all were connected via T1 circuits or better. We experienced no problems using the conventional logs it generated.
We have also had Bandwidthd running for about two years on a much smaller network. Initially there was an issue on this host with FreeBSD but an update resolved the bug that was found a few months ago in regards to the logs not being able to generate the stats after rebooting the server.
There have been other situations where we have used Bandwidthd and the results were always similar to the above examples. You can disable promiscuous mode on startup, otherwise I would probably leave it alone as well. The new version of bandwidthd gives the option to store the stats in PostgreSQL or in memory via text files. There is not much in the way of configuration unless you edit the src directly - (we added a line to include a css file and a few cosmetic changes) I think the config file is much simpler to use than MRTG, which was another issue the user had mentioned in the original email.
This is just the luck we have had with Bandwidthd... I have no idea if it has ever worked for anyone else.
More information about the freebsd-isp