Jakob, Hmm, both aliases are in the same network... This could be routing-related, as in that the ICMP replies could be all sent through eth0 (212.37.15.19), via the gateway for 130.240.192.145, because the routing table may be such that 130.240.192.145 is be reachable only through a gateway, which could be set up in the routing table to be reachable only through eth0 and not eth0:0). Maybe it could be something in the TCP/IP stack that has problems with two NIC's on the same subnet in one system (remark: your IP aliases are both on the same network, so there is no way to use network routing rules and for it to work well). Maybe the ICMP reply code doesn't look at the destination of the ICMP request, so it simply sends the ICMP reply through the NIC which it should be sent to according to the what the routing table says about the destination address for the ICMP reply (linux net guru's, your comments?). In order for the ICMP reply to be sent over eth0:0 in your case, I guess the ICMP reply code has to know that the ICMP request was directed to eth0:0 in the first place. But then again, if the routing table has no path to 130.240.192.145 via eth0:0, then the ICMP reply cannot reach it's destination via eth0:0... Since all hosts on the local 212.37.15.0 _are_ reachable through eth0:0, that could explain why the problem does not occur on the local network. I suggest trying your question on the linux-net or linux-tulip mailing lists. Post your netstat -r -n too, it may help. linux-tulip at [EMAIL PROTECTED] (Mr Becker's (the linux nic guy) list about the tulip cards). Or, maybe linux-net at vger. Jelle. On Sun, 11 Jul 1999, Jakob Sandgren wrote: > I have encountered a very strange problem with IP-aliasing and the > D-Link DFE-500TX nic. This seems very strange to me since I guess that > the IP aliasing code does not involve the nic driver(?). > > The problem has been verified with two DFE-500TX REV-E1 cards and the > following drivers: > tulip.c:v0.91e 5/27/99 > tulip.c:v0.91 4/14/99 > de4x5.c:V0.544 1999/5/8 > (All with 2.2.9) > This was tested in a SMP-box. > > This has also been tested with 2.2.10-ac10 on a UP-box with > "tulip.c:v0.91 4/14/9". > > * Everything works ok with the exact same setup, but with a 3c900 nic. * > > The strangest(?) part of this problem is that ip-aliasing work when > accessing the (virtual) host from the local network. But it's > impossible to access the (virtual) host from other networks. Once or > twice have I received replyes to ping request, but that's it. > > It looks more like a routing problem then a nic-driver problem. But > everything works fine with the 3c900 nic. > > > Some additional information (exploit): > > [root@echo log]# ifconfig eth0:0 212.37.15.17 > [root@echo log]# ifconfig > eth0 Link encap:Ethernet HWaddr 00:80:C8:F7:0F:DA > inet addr:212.37.15.19 Bcast:212.37.15.255 > Mask:255.255.255.0 > UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 > RX packets:21704 errors:0 dropped:0 overruns:0 frame:0 > TX packets:14769 errors:0 dropped:0 overruns:0 carrier:0 > collisions:24 txqueuelen:100 > Interrupt:9 Base address:0xdc80 > > eth0:0 Link encap:Ethernet HWaddr 00:80:C8:F7:0F:DA > inet addr:212.37.15.17 Bcast:212.37.15.255 > Mask:255.255.255.0 > UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 > Interrupt:9 Base address:0xdc80 > > lo Link encap:Local Loopback > inet addr:127.0.0.1 Mask:255.0.0.0 > UP LOOPBACK RUNNING MTU:3924 Metric:1 > RX packets:51 errors:0 dropped:0 overruns:0 frame:0 > TX packets:51 errors:0 dropped:0 overruns:0 carrier:0 > collisions:0 txqueuelen:0 > > [root@echo log]# tcpdump -n -i eth0 host 212.37.15.17 > tcpdump: listening on eth0 > 19:35:12.056232 212.37.15.14 > 212.37.15.17: icmp: echo request > 19:35:12.056259 212.37.15.17 > 212.37.15.14: icmp: echo reply > 19:35:13.050520 212.37.15.14 > 212.37.15.17: icmp: echo request > 19:35:13.050548 212.37.15.17 > 212.37.15.14: icmp: echo reply > 19:35:14.050518 212.37.15.14 > 212.37.15.17: icmp: echo request > 19:35:14.050540 212.37.15.17 > 212.37.15.14: icmp: echo reply > 19:35:17.049072 arp who-has 212.37.15.17 tell 212.37.15.14 > 19:35:17.049105 arp reply 212.37.15.17 is-at 0:80:c8:f7:f:da > 19:35:18.013666 130.240.192.145 > 212.37.15.17: icmp: echo request > 19:35:19.002895 130.240.192.145 > 212.37.15.17: icmp: echo request > 19:35:20.001946 130.240.192.145 > 212.37.15.17: icmp: echo request > 19:35:21.003478 130.240.192.145 > 212.37.15.17: icmp: echo request > > ...... > The tcpdump shows that the machine answers a ping from a machine on > the _same_ subnet. But when it receives a ping from a machine on a > different subnet, it's ignored. > > Anyone who has an idea about what this problem? > > /Jakob > - -- > > Please read the FAQ at http://www.tux.org/lkml/ > - To unsubscribe from this list: send the line "unsubscribe linux-net" in the body of a message to [EMAIL PROTECTED]
