I am asking this question again since I never got any solution. Additionally, I have more relevant information to this problem. I've a RHEL5 on a SunFire Server connecting through a LinkSys SRW2024 (24-port) switch. The switch is brand-new out of the box so it's settings should be set to default, unless a gremlin got in there at the factory while it was being shrink-wrapped in plastic. I am using the same subnet for the server as the switch. All the link lights are on and a double-checked by issuing the "ethtool eth0" command, says that the link was detected. I am not running DNS, DHCP, or anything else on this Linux server; I am entering in static IPs for the time being. I used 192.168.1.1 as the static IP address with a netmask of 255.255.255.0 and the gateway address left blank. I use either the GUI-driven network manager to enable the NIC card, or an eth0 up and an eth0 down. I am able to ping the localhost (127.0.0.1) as well as 192.168.1.1. However, I am unable to ping the switch, or any of the network client workstation. I get the classic "destination host unreachable" error. Someone suggested I try a crossover cable even though most of these switches have auto-uplink built in. The crossover cable didn't make a difference. I also confirmed that auto-link is running and that I could simply use a patch cable. I tried connecting workstation clients directly to the server, (using 192.168.1.2 as the client address) instead of through the switch. The server nor the workstation could see one another. I've tried setting a default gateway and that doesn't make a difference. I have also disabled SE Linux and the firewall. So iptables and filtering should not be an issue. I've tried telnetting (also manually specifying port 25) and I get:
"telnet: connect to address <address>: No route to host
telnet: Unable to connect to remote host: No route to host"
Upon seeing this, I figured "Maybe the switch is screwed up" even though it is brand new out of the box. I was able to see both client workstations THROUGH the switch. Not only that, I was able to use the LinkSys configuration webview from either client. So the only machine that cannot see anything else is the server. I checked the routing tables via netstat and route and I get the following (abridged output):
"192.168.1.0 * 255.255.255.0
169.254.0.0 * 255.255.0.0"
I again checked eth0 and the loopback device with "ifconfig eth0" and "ifconfig lo". It shows things as I would expect. I decided to try traceroute. Traceroute can't see a thing from the server. It can't get to the switch or to the client whether I have the client plugged in directly or not. Next, I decided to try nmap from both the clients AND the server. If I issue "nmap 192.168.1.0/24" from the client, it sees everything on the network. However, it says that all 1680 ports on the server (192.168.1.1) are being filtered. If I issue that same command from the server, it says that is is filtering only 1671 ports, with 9 open ports running. What the heck?! I have the firewall and iptables turned off. I don't understand. I've restarted xinetd with no change in behavior. I must admit that I am not a network expert but I have played around enough with a typical Linux workstation/server to administer some of the basics. Have I configured something wrong? What am I missing? Please help ASAP! I'm pulling my hair out!
