On Sat, 20 Jan 2007 23:26:54 -0600 "[EMAIL PROTECTED]" <[EMAIL PROTECTED]> wrote:
> On 1/20/07, Jochen Schulz <[EMAIL PROTECTED]> wrote: > > Bill Moseley: > > > > > > How would you go about tracking down the process that is eating up all > > > the bandwidth? > > > > Take a look at netstat-nat on the NATting machine iftop for the machine > > using the bandwidth. > > Unfortunately, iftop doesn't display usage per-process. Here are the > two solutions I've found: > > 1. atop, with the atopcnt kernel patch. Unfortunately, > kernel-patch-atopcnt doesn't apply against 2.6.18, so unless you use > 2.6.13 or 2.6.8, this doesn't help. (There's an RC bug filed against > that package, and hopefully it will be updated for Etch.) > > 2. A combination of iftop to find the port of the saturating > connection, and $(netstat -anp | grep portnumber) to find the process > that's using that port. > > There are several other bandwidth-monitoring apps, but I haven't yet > found the right one to show the usage per-process (other than atop). Nethogs [0] claims to display bandwidth per process. Celejar [0] http://nethogs.sourceforge.net/ -- ssuds.sourceforge.net - Home of Ssuds and Ssudg, a Simple Sudoku Solver and Generator -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
