Topic: internal IP connections

[Mikael]

Hello,

I just installed the e-smith 3.1 server on a lowly P166 box and even managed to get the integrated AMD 79C790 nic (pcnet32) working (though not officially supported) as eth1. The other nic is a supported Realtek 8029 (ne2k-pci) as eth0

The problem: when I connect this as a dedicated server & gateway between a cable modem and a lan hub, the connection is established perfectly from the gateway to internet addresses, yet I cannot get any connection between the firewall and other pc's within the lan. E.g. ping & telnet fails from the gateway to machines within a lan, as well as the opposite. Nor can i get
any gateway functionality...

================================================================
PING 192.168.1.1 (192.168.1.1) from 192.168.1.2 : 56 data bytes
From e-smith.mycompany.xxx (192.168.1.2): Destination Host Unreachable

From e-smith.mycompany.xxx (192.168.1.2): Destination Host Unreachable

--- 192.168.1.1 ping statistics ---
17 packets transmitted, 0 packets received, +15 errors, 100% packet loss

eth0   Link encap:Ethernet  HWaddr 52:54:AB:DD:38:94  
    inet addr:192.168.1.2 Bcast:192.168.1.255 Mask:255.255.255.0
    EtherTalk Phase 2 addr:65280/145
    UP BROADCAST RUNNING MULTICAST  MTU:1500  Metric:1
    RX packets:0 errors:0 dropped:0 overruns:0 frame:0
    TX packets:0 errors:111 dropped:0 overruns:0 carrier:0
    collisions:0

eth1      Link encap:Ethernet  HWaddr 00:00:4B:30:9C:74  
  inet addr:212.90.65.206 Bcast:212.90.65.255 Mask:255.255.255.0
  UP BROADCAST NOTRAILERS RUNNING  MTU:1500  Metric:1
  RX packets:97 errors:1 dropped:0 overruns:0 frame:2
  TX packets:16 errors:22 dropped:0 overruns:0 carrier:22
  collisions:0

lo        Link encap:Local Loopback  
          inet addr:127.0.0.1  Mask:255.0.0.0
          EtherTalk Phase 2 addr:0/0
          UP LOOPBACK RUNNING  MTU:3924  Metric:1
          RX packets:34 errors:0 dropped:0 overruns:0 frame:0
          TX packets:34 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0

Kernel IP routing table
Destination   Gateway    Genmask        Flags Metric Ref  Use Iface
192.168.1.2   0.0.0.0   255.255.255.255 UH    0      0    0 eth0
212.90.65.0   0.0.0.0   255.255.255.0   U     0      0    0 eth1
192.168.1.0   0.0.0.0   255.255.255.0   U     0      0    0 eth0
127.0.0.0     0.0.0.0   255.0.0.0       U     0      0    0 lo

================================================================
remarks:
** I changed the e-smith gateway internal IP from 192.168.1.1 to 192.168.1.2 since 192.168.1.1 was already reserved and all the machines within lan are already configured to use 192.168.1.2 as a gateway...
** I'm pretty sure there was also a a default gateway line more or less like:
default    212.90.65.1   0.0.0.0   UG    1   0    10  eth1
on the bottom of the routing table. That seems to have disappeared overnight, perhaps because it has been disconnected  from the cable modem.

Any ideas what I should start looking at in order to get at least the basic functionality on?

TIA,

       Mikael

[Roald Opsahl]

My best guesses, for what they are worth:

You say you changed the e-smith IP from 192.168.1.1 to 192.168.1.2 because
192.168.1.1 was reserved.  Reserved by whom for what?  Did  you already have a
192.168.1.1 machine before you installed e-smith? If so, is the ping example in your
message from the e-smith machine (192.168.1.2) trying to ping the machine you
already had from before (192.168.1.1)?

Anyhow, are you sure the wiring on the 192.168 net is good from the e-smith machine to the hub? According to the printout of your ifconfig, there hasn't
been any packages sent or received on eth0 since the card was activated. A bad patch cable, a cross-connected cable or a connection to an uplink port (marked
by X) would do you in.

Two alternate ways to test this:

* If you have two other machines connected to the hub, say 192.168.1.1 and 192.168.1.3 and they can ping each other, I would move the cable from
192.168.1.3 over to 192.168.1.2 (that's your e-smith right?) and then reboot
e-smith and try ping between xx.1 and xx.2.

* Swap the functions of the two cards on the e-smith, swap cables and IP
addresses between eth0 and eth1, if the problem remains the same I would really check the internal wirinf, if the problem follows the card I would review the HW and configuration including IRQ conflicts.

[Mikael]

Hi,

Thank you for your response. I realized right after the submit that I may have left out a few importat details, it's just that the post was getting a bit long even without them well here we go:
- Yes, I already had a 192.168.1.1 machine before I installed e-smith. Reserved by a NT server and changing it's IP would result some major network reconfiguration. Another important matter is, around 6 machines inside the lan are already configured to use 192.168.1.2 as default gateway.
- Yes, the ping was from e-smith (192.168.1.2) to (the previously existing) machine (192.168.1.1). I also tried from     e-smith to 192.168.1.3 and 192.168.1.4 (both running machines, all 3 of them are able to ping each other inside the lan) yet no reply
Furthermore:
- Both link leds - of the hub port and the one in the NIC are on - which imho rules out possibility of uplink port or an x-cable
- The card works flawlessly in another existing Red Hat 6.0 installation
- I don't see any problems at boot time during initializiation of eth0 (nor eth1)
- I tried another cable, and also the original cable works with another machine.
...all of which together would make me doubt a hardware failure, including IRQ conflicts. I do wonder if the problem is caused by the changed IP address of e-smith.

0f course I could still try to cross-assign eth1 and eth0 but other than that I'm afraid I'm running out of ideas...

brgds  

       Mikael

[Roald Opsahl]

Is must have been late last night for both of us. Anyhow, I would rule out the
change of IP as the reason, at least if you have rebooted since you made the changes.

Not that I'm so strong on this, but the number of errors in your ifconfig output
is too high and should be taken as a serious problem: for eth0 you have this for transmit:
TX packets:0 errors:111 dropped:0 overruns:0 carrier:0
and for eth1:
RX packets:97 errors:1 dropped:0 overruns:0 frame:2
TX packets:16 errors:22 dropped:0 overruns:0 carrier:22

I would be seriously worried about the percentage of errors, 111 of 111 is 100%
on eth0, 22 of 38 is close to 60% on eth1, not exactly what we call "perfect".
Clearly something is wrong, especially since you have your errors on transmit, I
think it is acceptable to see some errors on receive, but never on transmit. Here
are numbers from one of my machines:

  RX packets:21541862 errors:221 dropped:0 overruns:0 frame:221
  TX packets:3010013 errors:0 dropped:0 overruns:0 carrier:0
     
Trusting that your cables etc are good and given that you say that the RealTec
card works well in another machine/configuration I lean towards a conflict between the two cards or with some other HW in the machine.

I would still try to run with one card at the time , though the integrated card
is of course a bit difficult to remove. Is this a Compaq machine? If so, can't you
disable the NIC through the Setup/Diagnostic program?

Other relevant questions might be:

* What IRQ and such are you using for these two cards?
* What do you get from cat /proc/pci /proc/interrupts /proc/ioports?