|
Neohapsis is currently accepting applications for employment. For more information, please visit our website www.neohapsis.com or email hr@neohapsis.com |
From: Michael Curtis (mdc
ihug.co.nz)Date: Mon Jul 01 2002 - 17:59:45 CDT
Can anyone help me? :-)
I have installed OpenBSD 3.1 on an old (Pentium 133Mhz) dual-homed Compaq
machine with one onboard 10baseT NIC and a 1996 vintage PCI Compaq
Netelligent 100baseT card. Both these NICs use the tl (ThunderLAN) driver.
The onboard NIC is connected to an ADSL router on 192.168.240.0/24. This
connection works fine. The PCI NIC is connected to the office switch, and
squid/dhcpd/named are run as services to the rest of the network.
I have two problems with this machine. I am running the 3.xxx ISC dhcpd
because I need ddns updates and I am also using BIND9 from the ports tree for
a similar reason. The problem is this - one machine, a 10baseT machine,
which DOES NOT use dhcp (the operating system does't support it), can
connect, browse the internet generally talk to the router (named shark).
Another machine, a Windows 2000 Pro laptop (100baseT NIC), which is using
dhcp, cannot obtain an address from shark, and all pings, etc (obviously)
fail. However, swapping the switch for a hub and restarting shark causes
everything to work.
So, from this, I guess that one of the following situations is true:
1. OpenBSD is supposed to 'autoselect' the speed. If the selection is
failing and it's falling back to 10baseT, then any communication >10Mb/s will
fail as the OS is not expecting this. However, the NIC and switch both show
100baseT negotiation as succesful, so this does not seem likely.
2. For some reason, the dhcp broadcasts are not making it through the
switch. I can see that there is activity, because all the switch's activity
lights flash simultaneously when a dhcp-enabled device powers up. However, I
am concerned that the broadcasts from shark are not reaching the device.
Currently, no explicit broadcast address is specified, so hostname.tl0 reads:
inet 192.168.4.1 255.255.255.0
Could anyone shed any light on this?
The other problem is that I am trying to configure a redirect rule to point
to squid, so it acts as a transparent proxy. squid works as a traditional
proxy (on port 3128), and I have configured it to work as an 'httpd
accelerator'. But when I configure NAT to redirect to squid, all http
requests fail:
rdr on tl0 from any to any port 80 -> 192.168.4.1 port 80
I have tried these combinations: dest address 127.0.0.1, ports 80 and 3128,
and dest 192.168.4.1 port 80. No luck. What's going on?
This machine is the 'blueprint' for a set of four throughout the company, so
I really want to get all the bugs ironed out.
-- R ### Michael Curtis RISC OS fanatic I # # Electronics enthusiast S # # C # # Software and Acorn Users Waikato : O ######## http://homepages.ihug.co.nz/~jamjars/ S # Pidgin English for Sovereignty : Fella belong to Mrs Queen!
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]