Johnny Ljunggren

I know it is possible to route the whole net. And it is correct that parts of the ip range is already in use, that is why I added the 192.168.2.0 route in the example. There are only a few networks in use alread, and they are all below 192.168.100.0. More schematic: <IT servers> <-----> <IT Cloud> <----> <IT Switch> <-------> LAB Switch <------> LAB machines I want the lab machines (192.168.100+ ) to access a couple of IT servers. LAB machines have default gw through LAB switch. LAB switch has default gw through IT Switch. The IT Switch connects to the IT servers through the IT cloud - which consists of any numbers of routers/routing switches. My idea was to set gw on all the different IT switches so that traffic end up at the lab switch, unless specifically routed elsewhere on any of the switches. Hmmm, not sure if this made things any clearer though :)

Hello Is it possible to route C nets? I'm trying to get the IT department to add some routes so that our lab machines get access to a server in another network segment. However they are refusing and telling me they need to add routes for this one by one, and find it to be too much work. I want all networks 192.168.X.0 where X is 100-254 to go one way and the others to go wherever they need. Are they right that C nets cannot be routed like ip route 192.168.2.0 255.255.255.0 192.168.1.254 ip route 192.168.0.0 255.255.0.0 192.168.1.253 or something to that effect? regards Johnny

Thanks for the input. I had already tried this, but apparently failed to mentioned this in my previous messages. I think I've tried most things I normally do when trouble shooting stuff like this. This time it seems to be a Mandriva fault, not a configuration fault. ifconfig ethX <ip address> gives: SIOCSIFADDR: No such device ..... When using the network panel (gui style) the nic is listed, but when I click Configure I get: Unable to find network interface for selected device (using e1000e driver)

e1000e 112236 0

lspci -vv on Mandriva 2009.1 showed 'Kernel Modules: e1000e' but there was no line with kernel drivers in use, as with the other working NIC's. (output attached at the end) Another weird thing is that /etc/udev/rules.d/70-persistent_net.rules show a MAC address for this device, but it is not the correct one - 00:13:20.F5:F9:32 vs 00:01:29:21:2A:BF. To add to this when I boot Ubuntu the mac address attached to this nic is 00:00:08:00:02:88 - obviously wrong, but working! - I can set the ip address and ping the attached switch. Ubuntu show both Kernel Modules and Kernel drivers in use as e1000e. # cat /sys/module/e1000e/version show 0.3.3.3-k6 on both Mandriva and Ubuntu. lspci output on Mandriva 2009.1: 00:19.0 Ethernet controller: Intel Corporation 82566DC Gigabit Network Connection (rev 02) Subsystem: Intel Corporation Device 0000 Control: I/O+ Mem+ BusMaster- SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx- Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx- Interrupt: pin A routed to IRQ 20 Region 0: Memory at fdfc0000 (32-bit, non-prefetchable) Region 1: Memory at fdfff000 (32-bit, non-prefetchable) Region 2: I/O ports at ff00 Capabilities: [c8] Power Management version 2 Flags: PMEClk- DSI+ D1- D2- AuxCurrent=0mA PME(D0+,D1-,D2-,D3hot+,D3cold+) Status: D0 PME-Enable- DSel=0 DScale=1 PME- Capabilities: [d0] MSI: Mask- 64bit+ Count=1/1 Enable- Address: 00000000fee0f00c Data: 41a1 Kernel modules: e1000e lspci output on Ubuntu 9.04: 00:19.0 Ethernet controller: Intel Corporation 82566DC Gigabit Network Connection (rev 02) Subsystem: Intel Corporation Device 0000 Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR- FastB2B- DisINTx+ Status: Cap+ 66MHz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR- <PERR- INTx- Latency: 0 Interrupt: pin A routed to IRQ 2299 Region 0: Memory at fdfc0000 (32-bit, non-prefetchable) Region 1: Memory at fdfff000 (32-bit, non-prefetchable) Region 2: I/O ports at ff00 Capabilities: [c8] Power Management version 2 Flags: PMEClk- DSI+ D1- D2- AuxCurrent=0mA PME(D0+,D1-,D2-,D3hot+,D3cold+) Status: D0 PME-Enable- DSel=0 DScale=1 PME- Capabilities: [d0] Message Signalled Interrupts: Mask- 64bit+ Queue=0/0 Enable+ Address: 00000000fee0f00c Data: 41c1 Kernel driver in use: e1000e Kernel modules: e1000e

I've got a Intel Q965 System board equipped with a Marvell 88E805 and a Intel 82566DM onboard ethernet controller. The Marvell is configured as it should, but I am unable to get the Intel one configured. Running lshw -C network shows a Intel 82566DC (not DM) as UNCLAIMED and thus no driver or ethernet device connected. From what I gather this NIC should use the e1000e driver, but even if I add alias eth0 e1000e to modprobe.conf the nic remains unconfigured (have tried other ethX as well). Interestingly Ubuntu 9.04 works, so it seems there is something wrong with the hardware detection setup on Mandriva. I have tested this on Mandriva 2008.0 and 2009.1 and they show the same thing. I've also downloaded the e1000e source code and built and installed the driver, but no joy. By cat'ing the correct file in /proc I can see that the latest e1000e is loaded. Any pointers? regards Johnny

ianw1974 Thanks for your input. Unfortunately your description is the same as mine, and it still fails. xhost + will allow all machines, while xhost <ip address> only allows that one machine.

This is a really old machine that is on the verge to collapse, but until then we have to cope with what is on it.....

Me too, and it is quite annoying...

I forgot to add that ssh is not an option

Hello I've got an old machine running a specialized program that I want displayed on my computer. This has worked before and also works from another machine. These are the steps I do: 1. On my machine: xhost + 2. telnet oldmachine 3. setenv DISPLAY <my machine>:0.0 4. xclock Error: Can't open display: <my machine>:0.0 Doing the exact same steps from another machine works, so there must be something odd with my setup.

Thanks for the quick reply The systems run on their own dedicated network and there are usually around 10-15 machines so I'm not too concerned about network problems. It has worked so far but will it create problems in this situation as well? The rpm's in question are inhouse rpm's and only about ten different packages so I'm not sure what I will gain by using urpmi. I'd be happy to learn though. I'll see how much work it will be to use wget before installing. Thanks

hello I have set up some systems that will install RPM's over http from a central server The command I use is rpm -Uvh http://server/rpm/somepackage.rpm After the installation is finished I always get this warning (different numbers though): warning: u 0x8003faa0 ctrl 0x8004e0d0 nrefs != 0 Does anyone know what it means and possibly how to get rid of it? Johnny

Thanks for the input! Hmm, that's odd since both of the machines I'm working on come with 2GiB of RAM. free shows the same thing on both. No extra swap either. There is also a laptop-kernel included so there have to be other rationals as well. What parts of the installation is responsible for this selection? I'd assume it is possible to read the source code to figure this out.

Hello I've got two computers that essentially have the same hardware, but there are some small differences. When installing, one of them will get the desktop kernel and one will get the server kernel. What is the rationale for choosing one or the other? I assume it is hardware related but what parts? Johnny