| View previous topic :: View next topic |
| Author |
Message |
your_WooDness
Tux's lil' helper
Joined: 25 Oct 2007
Posts: 77
|
 Posted: Wed Jan 21, 2009 9:09 pm Post subject: [SOLVED] one MAC address for several IPs |
 |
|
Hi out there,
I am having a very strange problem with Intel NICs and their MAC addresses.
For example a server with 2 Intel NICs: if you enter a "ifconfig" you get two IP addresses and two MAC addresses. Everything as it should be. If I e.g. ping both IP addresses from another host and then check the arp-table, I will get one and the same MAC address for both IP addresses. I hope the folllowing snippit will make it clear...
[edit]
machine A has the IP 10.0.0.10
machine B has the IPs 10.0.0.20 and 10.0.0.21
[/edit]
Locally on B the "ifconfig" output of the machine with two intel NICs shows this
| Code: |
rose src # ifconfig
eth0 Link encap:Ethernet HWaddr de:ad:be:ef:22:78
inet addr:10.0.0.20 Bcast:10.0.0.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:873 errors:0 dropped:0 overruns:0 frame:0
TX packets:701 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:84366 (82.3 KiB) TX bytes:161631 (157.8 KiB)
eth1 Link encap:Ethernet HWaddr de:ad:be:ef:22:79
inet addr:10.0.0.21 Bcast:10.0.0.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:1 errors:0 dropped:0 overruns:0 frame:0
TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:42 (42.0 B) TX bytes:0 (0.0 B)
|
When I now ping these two addresses on B from A and have a look at my arp-table of A:
| Code: |
golem wood # ping 10.0.0.20
PING 10.0.0.20 (10.0.0.20) 56(84) bytes of data.
64 bytes from 10.0.0.20: icmp_seq=1 ttl=64 time=0.443 ms
golem wood # ping 10.0.0.21
PING 10.0.0.21 (10.0.0.21) 56(84) bytes of data.
64 bytes from 10.0.0.21: icmp_seq=1 ttl=64 time=2.00 ms
golem wood # arp
Address HWtype HWaddress Flags Mask Iface
localhost ether 00:13:49:07:69:d7 C eth0
10.0.0.20 ether de:ad:be:ef:22:78 C br0
10.0.0.21 ether de:ad:be:ef:22:78 C br0
*** this example is kvm based, but also appears on vmware and bare metal ***
|
...I see that I have the same MAC address for 10.0.0.20 and 10.0.0.21 in the arp-table of A. Well, this can lead to a very strange behaviour in the network, I dont't think I have to go on... =0)
The first time I have encountered this problem I was troubleshooting an IPMI problem for a customer of my company and I thought it would be related to IPMI. But it's not, 'cause I have another server completely without IPMI (2 onboard NICs and a QuadPort Intel NIC) which shows the same problem: one MAC address for several IP addresses.
I then doublechecked this within a vmware based virtual machine and a kvm based virtual machine, which both emulate a e1000 NIC. Both installations showed the same problem. One installation is a SuSE Enterprise 10 SP1 and the other one a gentoo installation. So I think the problem is rather related to the e1000 module than hardware related.
Has anyone seen this already and/or has a workaround for this?
WooD
Last edited by your_WooDness on Mon Jan 26, 2009 6:26 am; edited 2 times in total |
|
| Back to top |
|
 |
cyrillic
Watchman
Joined: 19 Feb 2003
Posts: 7313
Location: Groton, Massachusetts USA
|
| Posted: Wed Jan 21, 2009 9:27 pm Post subject: |
|
|
I am not sure having eth0 and eth1 on the same subnet accomplishes anything. Most likely, all the traffic is going through eth0, and eth1 is just taking up space.
What does "route -n" look like ? |
|
| Back to top |
|
|
linuxtuxhellsinki
l33t
Joined: 15 Nov 2004
Posts: 700
Location: Hellsinki
|
| Posted: Wed Jan 21, 2009 9:37 pm Post subject: Re: Intel NIC / e1000 module / one MAC address for several I |
|
|
| your_WooDness wrote: |
| Code: |
golem wood # arp
Address HWtype HWaddress Flags Mask Iface
localhost ether 00:13:49:07:69:d7 C eth0
10.0.0.20 ether de:ad:be:ef:22:78 C br0
10.0.0.21 ether de:ad:be:ef:22:78 C br0
*** this example is kvm based, but also appears on vmware and bare metal ***
|
|
Iface = br0 
_________________
1st use 'Search' & lastly add [Solved] to
the subject of your first post in the thread. |
|
| Back to top |
|
|
your_WooDness
Tux's lil' helper
Joined: 25 Oct 2007
Posts: 77
|
| Posted: Wed Jan 21, 2009 9:58 pm Post subject: |
|
|
Hi,
this is just an example and br0 does not really matter, 'cause I have seen this on br0, eth0 and eht1 (baremetal). This has absolutely nothing to do with a bridge config or bondage config. The NICs, either VM or baremetal, are just configured the normal way.
WooD |
|
| Back to top |
|
|
your_WooDness
Tux's lil' helper
Joined: 25 Oct 2007
Posts: 77
|
| Posted: Thu Jan 22, 2009 6:53 am Post subject: |
|
|
Hi there,
"route -n" gives the following:
| Code: |
golem wood # route -n
Kernel IP routing table
Destination Gateway Genmask Flags Metric Ref Use Iface
10.0.0.0 0.0.0.0 255.255.255.0 U 0 0 0 br0
192.168.1.0 0.0.0.0 255.255.255.0 U 0 0 0 eth0
127.0.0.0 127.0.0.1 255.0.0.0 UG 0 0 0 lo
0.0.0.0 192.168.1.1 0.0.0.0 UG 3 0 0 eth0
|
I had a closer look on this and tried several Intel e1000 modules, but the problem remains. Maybe I haven't found the correct one yet. =0)
Then I tried to add the correct MAC addresses statically in the arp-table of A via:
| Code: |
golem wood # cat /etc/ethers
10.0.0.20 de:ad:be:ef:22:78
10.0.0.21 de:ad:be:ef:22:79
golem wood # arp -f /etc/ethers
|
Then I had the correct MAC addresses in my arp-table on A, but the 10.0.0.21 address was unreachable. Setting the arp table back for 10.0.0.21 on the "wrong" MAC address (de:ad:be:ef:22:78 ), the 10.0.0.21 was reachable again.
Hm...strange. When you have a look at the activity LEDs of the two NICs of machine B only one is blinking and shows activity when you ping both IP addresses of machine B. So physically the traffic through the network must be directed through 1 NIC on machine B and somehow it must be redirected on machine B to the correct IP address.
Well, I think this should also affect the performance of the network connections when two or more IP addresses share one physical port, but I haven't made any performance tests yet.
I also disconnected the cable of the NIC with 10.0.0.20 / de:ad:be:ef:22:78, both IP addresses of machine B were not reachable anymore, although the NIC with 10.0.0.21 / de:ad:be:ef:22:79 showed a hardware link. If you disconnect the cable of NIC 10.0.0.21 / de:ad:be:ef:22:79, both IP addresses of machine B are still available.
WooD |
|
| Back to top |
|
|
your_WooDness
Tux's lil' helper
Joined: 25 Oct 2007
Posts: 77
|
| Posted: Thu Jan 22, 2009 8:37 pm Post subject: |
|
|
Hi,
just a little update. I have tried to set two different IP addresses for the NICs:
| Code: |
rose src # ifconfig
eth0 Link encap:Ethernet HWaddr de:ad:be:ef:22:78
inet addr:10.0.0.20 Bcast:10.0.0.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:873 errors:0 dropped:0 overruns:0 frame:0
TX packets:701 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:84366 (82.3 KiB) TX bytes:161631 (157.8 KiB)
eth1 Link encap:Ethernet HWaddr de:ad:be:ef:22:79
inet addr:192.168.1.100 Bcast:192.168.1.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:1 errors:0 dropped:0 overruns:0 frame:0
TX packets:0 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:42 (42.0 B) TX bytes:0 (0.0 B)
|
If I then ping both IP addresses and have a look at my arp table everything is OK:
| Code: |
bear wood # arp
Address HWtype HWaddress Flags Mask Iface
localhost ether 00:13:49:07:69:d7 C eth0
10.0.0.20 ether de:ad:be:ef:22:78 C br0
192.168.1.100 ether de:ad:be:ef:22:79 C br0
|
If I switch back to my initial IP setup 10.0.0.20 and 10.0.0.21 I will have the same MAC address again in the arp cache. Well, the problem seems only to occur if you have the same subnet on the NICs.
WooD |
|
| Back to top |
|
|
your_WooDness
Tux's lil' helper
Joined: 25 Oct 2007
Posts: 77
|
| Posted: Mon Jan 26, 2009 6:25 am Post subject: |
|
|
Hi there,
the problem I have reported in this thread is the linux arp flux problem. Haven't seen this before...
WooD |
|
| Back to top |
|
|
|
Kernel & Hardware
