Hi Ervin, On Thu, 2008-02-07 at 23:37 +0100, ervin wrote: > guidance for OpenBSD Bootstrap here -> http://www.ultradesic.com/?section=74
Which does not reveal an awful lot about specific trouble and possible errors.. In fact, I recently tarted up my own doc, which is a bit more verbose, especially in the error section (which is why I wrote it in the first place): http://stsx.xs4all.nl/www.stsx.org/openbsd/obsd-bootsoekris.html (Added a few lines about firewalling this morning). Thanks anyway, and Brian as well, Bill By the way, I think my initial error was a faulty "next-server", after that I went into a chain of overlapping fixes/new errors. It happens.. > On 06/02/2008, Brian Candler <[EMAIL PROTECTED]> wrote: > > On Wed, Feb 06, 2008 at 12:16:45PM +0100, Bill Maas wrote: > > > I've been running dhcpd/tftpd on OpenBSD 3.8-4.2 on a PC without > > > problems, for booting my Soekris boxens. However, when I run [almost] > > > the exact same config on a net4521, and try booting another Soekris box > > > off that one, TFTP times out. On the boot server, 'inetd -d' shows a > > > rapid re-execution of tcpd (or of tftpd without using TCP wrappers) when > > > the client tries to fetch the pxeboot image. The same happens by the way > > > when I try to fetch files manually using 'tftp' from Linux or OpenBSD, > > > but those implementations seem to have a reputation for stupidity and > > > brokenness. > > > > > > The really weird thing is that the same config works on a PC as boot > > > server, but not on a Soekris box. Anyone seen this before? > > > > No, but my suggestion would be to disable your OpenBSD pf temporarily and > > see if the problem goes away. > > > > TFTP is a pretty ugly protocol. An exchange looks like this: > > > > client UDP port XXX ------------> UDP port 69 server > > > > UDP port XXX <------------ UDP port YYY > > > > UDP port XXX ------------> UDP port YYY ...etc > > > > That is, the server choses a dynamic port for sending its replies from. It's > > not easy to firewall properly. > > > > Maybe the difference between your clients is that one happens to have chosen > > XXX=69 as its source port, but the other has chosen XXX=dynamic high port > > (which is the "normal" thing to do). tcpdump could confirm or refute that > > hypothesis. > > > > Regards, > > > > Brian. > > _______________________________________________ > > Soekris-tech mailing list > > [email protected] > > http://lists.soekris.com/mailman/listinfo/soekris-tech > > > > > -- SOD Against NLS _______________________________________________ Soekris-tech mailing list [email protected] http://lists.soekris.com/mailman/listinfo/soekris-tech
