Solved Intel 82579V Gigabit Ethernet Controller issue

diizzy: Thanks for your suggestion to use DHCP. I will try that a little later as I want to use DHCP as the last resort. I am getting this feeling that I may have mis-configured my network so want to investigate with the static IP.

Terry_Kennedy:
I have corrected ifconfig_em0 to -rxcsum -txcsum -tso and restarted the server. I ran the same tcpdump exercise again for a bit longer this time and I have attached the findings.

Can you ping any of those addresses (192.168.0.6, 192.168.0.7, or 192.168.1.1) successfully?
I can ping 192.168.1.1 successfully which is another router connected to an unmanaged switch. The switch in turn is connected to this TotoLink wireless router which is why I have changed the IP address on it. The switch also has connections to a LAN that has 192.168.1.x IPs. 192.168.0.6, 192.168.0.7 are addresses of two other desktops in the WLAN. I am unable to ping 192.168.0.6 or 192.168.0.7.
 

Attachments

  • tcpdump.txt
    7.6 KB · Views: 240
I have corrected ifconfig_em0 to -rxcsum -txcsum -tso and restarted the server. I ran the same tcpdump exercise again for a bit longer this time and I have attached the findings.

I can ping 192.168.1.1 successfully which is another router connected to an unmanaged switch. The switch in turn is connected to this TotoLink wireless router which is why I have changed the IP address on it. The switch also has connections to a LAN that has 192.168.1.x IPs. 192.168.0.6, 192.168.0.7 are addresses of two other desktops in the WLAN. I am unable to ping 192.168.0.6 or 192.168.0.7.
Ok, I see the ICMP echo requests in the tcpdump output, but you aren't receiving any replies.

There is something very strange with your LAN, however. Normally everything on a LAN should be with a single IP address / netmask range. In your case, you have 192.168.0.x/255.255.255.0 which is the network your FreeBSD box thinks it is attached to. But there is also a 192.168.1.x network, presumably also with a 255.255.255.0 mask. It is very strange that you can ping 192.168.1.1 (which is not on what the FreeBSD system thinks is your local LAN) but not be able to ping other systems on the 192.160.0.x LAN. They may be configured to not respond to ICMP echo requests, however.

I realize this is frustrating, but I think we're getting close. Please do # traceroute -n 192.168.1.1 after verifying you can still ping that address. Also, please post or PM me the /etc/rc.conf files from both this system and the FreeNAS 9.3 system which you earlier said was working (with a re(4) controller).
 
Yes hw.pci.realloc_bars as well as hint.agp.0.disabled=1 were added to /boot/loader.conf after googling for similar issues and certain people mentioned that their issue got sorted thereafter.

Now, I have commented out hw.pci.realloc_bars, hint.agp.0.disabled=1 and if_em_load="YES" and restarted the server.

PR 196501 is what refers to this and there was some specific hardware items in that that brought about a PCI bus issue. It's not an em(4) issue and those configuration options shouldn't be there unless you are in the same situation as the bug.

This is the part that catches my eye.
Code:
15:28:40.993325 IP myFreeBSDServer.54421 > 192.168.0.10.domain: 18519+ PTR? 7.0.168.192.in-addr.arpa. (42)
15:28:41.066140 IP 192.168.1.1.domain > myFreeBSDServer.54421: 18519 NXDomain* 0/1/0 (97)

I'm fairly certain you've got your networking mixed up. Typically, the UDP reply from a DNS query would have the source address of the closest interface. Unbound for example requires interface-automatic: yes to always return DNS requires with the address they were sent to rather the closest interface. I don't know what DNS software runs on that router but it gives me a strong suspicion that your server should be on 192.168.1.0/24.
 
I am beginning to feel that is what it is more likely than not.
I got your diagram, but I haven't received the IP configuration info for the FreeNAS box that works.

You are doing double NAT, which you generally want to avoid if possible. "Why" is a discussion for another time.

Without info on the FreeNAS config, I would suggest connecting the FreeNAS and FreeBSD systems directly to the unmanaged switch and putting them back in 192.168.1.x (don't forget to change the default route as well). That will leave 192.168.0.x as your wireless network.

Later on, if you want to get rid of the double NAT, you could configure your wireless router as an access point. That involves moving the connection to the switch (which should be the only wired Ethernet connection) to one of the LAN ports (right now it is in the WAN port) and disabling DHCP, etc. on the wireless router. You would also assign the wireless router an address on the 192.168.1.x network. Wireless clients will receive DHCP addresses from your main 192.168.1.1 router. At that point, nothing will be using 192.168.0.x.
 
Back
Top