network boot/dhcp issues??

[gtsupport]

Hi,

I haven't had any real breakthrough with the issues on my core/hybrid, so I thought I would tackle adding a MD, just to try and get a positive result somewhere!

Unfortunately this wasn't the case, and I need some advice on where to check.

I'm pretty sure that my network connections are working ok now, the core can connect to the internet without problems, and at one point I even had my windows laptop working on-line using the internal connection on the core, which gave it an IP address, just like it should.  I'm using 2 NICs as suggested, all good stuff.  Here's where it went pearshaped!

I have a zotag MAG which I setup to network boot.  Switched on, and it said all the right things about announcing ourself to the network or whatever, then started installing the diskless setup.  This went on for hours, and never ended.  At that point I had run out of play time, and had to move the core, and MD off of the dining room table, (SWMBO's orders).  So having given up on the network boot after a couple of hours, I switched everything off, and moved the core to its rightful place.  Everything is now connected back as it was, but now when I switch on the MD it tries to network boot, but after 1 minute gives up.

As I said, I'm pretty sure the internal network is configured ok, I've read heaps of threads going back to 2006, and some really useful ubuntu ones, but I'm not sure whats not working.  I typed "sudo tail -20 /var/log/daemon.log¦grep "dhcpd" and the result was:

(time & date) DHCPDISCOVER FROM (MD mac address) via eth1
(time & date) DHCPOFFER ON 192.168.80.2 to (MD mac address) via eth1

which seems to suggest to me that the DHCP service is running. (Confirmed by typing /etc/init.d/dhcp3-server status = Status of DHCP server: dhcpd3 is running)

So what must I do to get LinuxMCE to allow it to PXE Boot?

{insert animated smiley pulling hair out}

Cheers
Matt.

[gtsupport]

Oh, and please can somebody help with this unrelated question???

I seem to be spending all my time in a bash shell on the KDE desktop.  Whenever I try to get back to the Orbiter, it doesn't come back up.  I guess a screensaver or something has kicked in.

Is it possible to run a command prompt from anywhere inside LinuxMCE, or is there a way to restart the orbiter from the KDE desktop/bash shell?

Finally, what is the correct way to shut down the core?? Whenever I try, from either the orbiter screen, or the KDE desktop it just gives me a grey screen with a black mouse pointer....forever.  I have to hold the power button down to switch it off.  Perhaps this is linked to my other problem(s)?  If I CTRL + ALT + F(n) then I can select different TTYs but they won't allow any keyboard or mouse input.  Occasionally I get the Kubuntu logo with the blue bar going backwards, which looks like it should be shutting down, but then I get a black screen and the power-button-holding comes back into play.

Happy LinuxMCE days  

[purps]

Put a shortcut on your desktop with the command "/usr/pluto/bin/ActivateOrbiterFromKDE.sh", that will bring the orbiter back up.

I usually power down my core from the orbiter, has always worked fine. Try pressing the power button straight off to shutdown the core. Alternatively, run "sudo shutdown -h now" in a terminal.

Regarding your first post, it seems that you got quite far, and then got thwarted by something else. I would therefore suggest you remove any MDs through webadmin (DO NOT remote the core/hybrid MD), give everything a reset, and start again with your MD install. Remember you might have to select your video card if you get a blank screen http://wiki.linuxmce.org/index.php/AVWizard#What_to_do_if_A.2FV_Wizard_does_not_start

Cheers,
Matt.

[gtsupport]

Thanks again purps for helping out, those tips regarding restarting the orbiter will have saved me many core re-starts.

 I tried removing the MD in the web admin, it did warn me that it would ruin my system or words to that effect, but I did it anyway.  Then when I switched the MD on, again it didn't start booting, just sitting there with the cursor spinning round with "dhcp" which suggests to me that its waiting for an ip address.  Anyway I checked the daemon.log file again and this time the MD was allocated an ip address of 192.168.80.129 which is outside the linuxmce range, and suggests that the core thinks that this is a different computer, not a MD.

Anyway as it didn't work, I tried adding a new MD in the web admin, and entered the mac address of my MD.  After this it appeared in the MD section, and also was given a moon number (50).  Also the dhcp assigned it a 192.168.80.2 IP address which is what I'd expect.  However the MD still just sits there waiting as before.  Could this be a problem with the NIC??  The internal nic is the one assigned as eth1 and is the onboard one.  I'm wondering if it's somehow allowing traffic one way only??  Seems unlikely, but I don't really know what else to think.  Because my motherboard requires a driver for the Realtek RTL8111E NIC I downloaded the necessary driver, and installed it.  Is there any way that LinuxMCE is overriding this driver or something?

I've got a seperate gigabit switch between the MD and the core, it shows that both ports are active so probably my thinking is wrong.  Is there anything I can check, or try to make sure that the internal network is operating properly?

Cheers.

Matt.

[gtsupport]

Out of desperation I've been going through the dmesg log to see if there's anything amiss.  I found these entries, can anyone advise on the unknown mac address for the 2nd NIC..?  Plainly it has got one, as its listed here.

[    2.654377] r8169 Gigabit Ethernet driver 2.3LK-NAPI loaded
[    2.654390] r8169 0000:06:00.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16
[    2.654403] r8169 0000:06:00.0: setting latency timer to 64
[    2.654669] eth0: RTL8168b/8111b at 0xf8a2c000, 00:1f:1f:c0:c6:40, XID 38000000 IRQ 217
[    2.655748] r8169 Gigabit Ethernet driver 2.3LK-NAPI loaded
[    2.655756] r8169 0000:01:00.0: PCI INT A -> GSI 16 (level, low) -> IRQ 16
[    2.655767] r8169 0000:01:00.0: setting latency timer to 64
[    2.655787] r8169 0000:01:00.0: unknown MAC (2f200000)
[    2.656013] eth1: RTL8169 at 0xf8a3c000, 20:cf:30:ae:ee:e2, XID 2c200000 IRQ 216


Also what is strange is that the RTL8168 NIC is the onboard one, which as far as I know is setup as eth1, the PCI NIC is operating as eth0.  The only way I managed to get LinuxMCE to install is by using the PCI NIC connected to the router.  From then on its always worked like that, if I swap the cables over, then I lose internet connectivity from my core.

Later on in dmesg I see:-

[   36.714622] r8169: eth0: link up
[   36.714626] r8169: eth0: link up
[   41.042621] r8169: eth1: link down
[   41.042744] ADDRCONF(NETDEV_UP): eth1: link is not ready
[   47.292675] eth0: no IPv6 routers present

When I loaded Kubuntu I had to install new drivers for my onboard NIC which was r8168.ko  and I would have thought that this should have appeared somewhere, but I only see r8169 in the logs.  Is this something to worry about?

I've tried swapping the interfaces around in webadmin, but this just brings the internet connection down, and then I can't access the dcerouter either.  I think there is something amiss with the onboard nic, but how can I put it right?

Its becoming so frustrating, I only want 1 thing to work properly to give me a boost

[purps]

I tried removing the MD in the web admin, it did warn me that it would ruin my system or words to that effect, but I did it anyway.  Then when I switched the MD on, again it didn't start booting, just sitting there with the cursor spinning round with "dhcp" which suggests to me that its waiting for an ip address.

Sounds like you removed the core/hybrid MD, which I did say not to do in my last post. No idea if you can fix this, I would start again if I were you.
When you see these message "This could seriously screw up your system", do please take notice of it, it is there for a reason!

[gtsupport]

No, it was definately just the media director, it was allocated moon 50, so I know it wasn't the core/hybrid that I removed.  I did manage to get a bit more sense out of it by switching the network interfaces, then switching the patch cables over and rebooting.  I switched the MD on, and it got its ip address straight away, then the kubuntu logo appeared, the blue line filled up, then I have a screen filled with text....ending in:-

* Restarting OpenBSD Secure Shell server sshd

At this point both the MD stop working, and worryingly so does the core.  It won't respond to any keypresses, mouse movements, or gyration remote buttons.  So just when it looked like it was getting somewhere, back to square 1 again...

*Update*...

Just tried it again after 2 more restarts, and got the MD to the point where it should have started the AVWizard.

No keypresses will provoke any response, I've tried HDMI, and VGA outputs from the MD.  I can however ssh into the MD from the core, I tried running the AVWizard from there, but it responds with command not found, which is strange, because I can clearly see the command when I give the command ls in /usr/pluto/bin.

Just tried a soft reboot on the MD, and it froze the core again.   Grrrr.

[gtsupport]

Ok, so I now know that my onboard NIC is not working for most of the time.  Its very strange, as sometimes it works, then another problem causes me to have to reboot the machine, and after that it doesn't work.

I tried a fresh install from the 23388 DVD again, and the strange thing is that when kubuntu installs, the NIC doesn't work, but when I install the r8168 patch it works.  I have been browsing the web from the KDE desktop several times, and even after multiple reboots it still works.  I then installed LinuxMCE, and it installed fine, got all the updates and installation files from old-releases.ubuntu.blah which proves that the NIC is working, and there is a valid net connection.

The trouble starts once LinuxMCE has installed and restarted the machine.  From this point onwards the NIC won't work.  Obviously something is killing the driver, or overwriting it with a newer version or something.  If I go back and reinstall the patch again it makes no difference.  Where can I go and see whats happening?  I know that there is a seperate wiki page for getting the r8168 card working, but it only mentions version 710.  Can anyone tell me if its still valid for 810?

http://wiki.linuxmce.org/index.php/R8168

I'm hoping that once this is sorted, then all my network boot issues will be over ;o)

[purps]

Maybe this is the problem I am having as well http://forum.linuxmce.org/index.php?topic=11159.0

It didn't work at all whilst using the live CD or with the standard KDE desktop (had to plug the Internet into my working second NIC to do updates/installation). Then, I could have SWORN that it worked for a while after the LMCE installation (without installing patches or anything like that), but then stopped working the next day.

[gtsupport]

Eureeka!!!

Thanks again to Purps for showing me the way!

http://ubuntuforums.org/showthread.php?t=1091989

I followed this link and done as instructed.  Finally I got both NICs working.  It seemed that I had the same problem as Purps in that there were 2 rtl modules loaded, and I can only assume that they were in conflict somehow.  Anyway after blacklisting the r8169 module, updating, and rebooting, I managed to get internet connectivity on 1 NIC, and the MD was granted an IP address on the 2nd.  Just waiting for the Diskless_CreateTBZ.sh command to complete to see if the MD will netboot.

Feeling much better about all my hardware working with LinuxMCE now, I'll put that Windows7 DVD back in the drawer!!

[purps]

Did you have to install anything manually from Realtek, or did you just do the blacklist part?

[gtsupport]

I downloaded the r186-8.020.00 file from realtek.  It installs from an autorun.sh script once you've unzipped the tarball.

As I said, it worked straight away on a clean Kubuntu install, and only messed up when LinuxMCE installed.  I'm glad to say it still works, and I've begun the long road to getting it setup how I want it.

Hope you get yours working mate.

Matt.

[purps]

Well I am glad it is working for you.

Both of your cards must be using the same module i.e. the r8168 you've installed.

My problem is one card needs the r8168 module, and the other needs the r8169 module (which you have removed in your installation). I don't think there's any crime in having both installed, but there is some issue with the PCI IDs overlapping, hence there's conflict, and I have no idea how to sort it out.

I bought a new intel NIC on ebay, so hopefully that will work out the box and I can have two NICs once again.