Re: performance of libnet-based tools
Frédéric Raynal <[email protected]> Tue, 1 Mar 2005 20:23:18 +0100
| Newsgroups | gmane.comp.security.libnet |
|---|---|
| Message-ID | <[email protected]> |
On Fri, Feb 25, 2005 at 11:07:12PM -0800, Steven Alexander Jr. wrote: > I've been experimenting with a few different > libnet-based tools such as Packit and ISIC. If I try > to generate a high-volume of traffic using these > tools, with various options and protocols, the error > rate is extremely high. ISIC says: > > Failed to send packet: No buffer space available > > The synflood tool that comes as a libnet sample works > for several seconds and begins reporting: > > libnet_write: libnet_write_raw_ipv4(): -1 bytes > written (Host is down) > > I'm not sure where the problem lies and I would love > to hear any suggestions. I'm using a 3.0Ghz P4 with > 1.5G RAM and a $20 Linksys FA311 card on FreeBSD 5.3 > and a 1.3GHZ AMD laptop with 512MB RAM and a built-in > Realtek RTL8139C+ card on FreeBSD 4.9. > > Do I need to tune the OS? Do I need a better network > card? > > I'd like to be able to perform various stress testing > exercises on my test network. > I'm also very interested in those tests. Can you tell us more please? Now, I'll try to answer some of your questions. "Failed to send packet: No buffer space available" : this may be caused because the driver's buffer is full. " libnet_write: libnet_write_raw_ipv4(): -1 bytes written (Host is down)" : here, that may be because the remote host is really down ? I mean some OS do not feel very well under synflood. That is just guesses ... I may be mistaken ... A+ Fred Raynal