Setting up remote boot nodes help!!

[PCZ]

The nodes are nforce2 chipsets with nforce onboard nics.
The server is red hat 9
I am going to install the RPM's from ltsp.org.

I have a nasty feeling that the nforce nics won't work somebody please tell me that they will.

Brian

[Dyyryath]

I'm afraid not. :(

It looks like only the nForce3 Pro (http://www.nvidia.com/page/nforce3.html) boards have nVidia network chipsets capable of netbooting as seen here (http://www.nvidia.com/object/feature_network.html).

I suspect that you're going to have to buy some cheap PCI network cards & use them to net boot. :cry:

Of course, if you find out differently and make it work, be sure to let us know!

[IronBits]

Originally posted by Dyyryath
I'm afraid not. :(
I suspect that you're going to have to buy some cheap PCI network cards & use them to net boot. :cry: That's exactly what I had to do using SuSe and Mandrake on my Asus A7N8X boards :bang:

[Dyyryath]

Were you trying to net boot, IB? There are Linux drivers for the nForce, you just have to get them from nVidia (most distros don't come with them). The chipset just doesn't (that I can see) have net boot capability.

[PCZ]

What about the via nics built in to the km266 chipset.
Do they net boot ?

[magnav0x]

PCZ, I'm not sure whether the PXE depends on the motherboards chipset, the motherboard's bios vendor, or the onboard nic chipset. Hopefully someone here knows, because I'd like to know :help:

[IronBits]

I'll let you know when I get my hands on one. ;)
Some of the forums I found digging thru google, indicate the VT6103 is PXE if the BIOS allows you to choose it as a boot device.
So I went thru the Biostar BIOS screens and yup, it shows it as a selectable boot up device.

Dyyryath, an add-on driver during install, which leaves me out in the cold with *nix boxen.
If it don't come up right the first time, then I'm toast. Either I re-install and hope for the best, or move on to the next *nix version.
I have two nics installed in the Mandrake Server with the onboard turned off, because that's how it worked ;)
I remember jack squat about unix, other than ps, ls and rm <LOL>

[Paratima]

Originally posted by IronBits
II remember jack squat about unix, other than ps, ls and rm <LOL> There's more? :confused: :jester:

[magnav0x]

IB, if you want an overview of what you are doing and don't have dyyr to help you out, you may try looking over at the Monster Farm HOWTO @ http://ws9.jobnegotiator.com/html/netboot.html

It covers diskless booting via PXE

[PCZ]

Maybe they can do it after all.

This is what I see if I select Lan as a boot option
Aopen AK79G-1394 Nforce 2

NVIDIA Boot Agent, PXE-2.0 (build 082 v1.82)
Copyright (C) 2001 NVIDIA Corporation

[IronBits]

That looks promising :thumbs:
Does that mean you are going to add MORE boxen?? :scared: :D

[PCZ]

Maybe
I won't promise anything just yet.

If I do add anymore nodes I have decided they will have to be remote boot.
I installed red hat 9 without the GUI and working at the command line with an alien OS is tough. I didn't install a TFTP server so I am a bit screwed now.
I don't know how to add it from the CD.

[magnav0x]

PCZ, you'll have to mount the CD's one by one and find the tftp rpm.

Insert CD1:

mount /dev/cdrom /mnt/cdrom
cd /mnt/cdrom/RedHat/RPMS
ls *tftp*

If it says the file doesn't exist then unmount the cdrom:

cd /mnt
umount /dev/cdrom

And repeat the first process with CD2 and CD3 or until you find the tftp file. It'll probably be called something along the lines of tftp-server-0.32-1.i386.rpm

When you find the file just type in:

rpm -i tftp-server-0.32-1.i386.rpm

Of course replace tftp-server-0.32-1.i386.rpm with your version's filename. Hope that helps.

[PCZ]

magnav0x
Thanks for the Info.

I had already worked out how to mount the cdrom but I couldn't find the tftp rpm on any of the CD's. I wasn't using LS properly I expect.
I worked out that eject releases the CD as long as you are back at the root. That had me going for a while.
I got the TFTP RPM from the web.

I keep getting an error. INIT: "x" respawning too fast: disabled for 5 mins.
It doesn't appear to be doing any harm but it is annoying.

[Dyyryath]

It means that your X server config is boogered. If you don't care about X, then edit the file '/etc/inittab' and change the following line:

id:5:initdefault:

to this:

id:3:initdefault:

That will start you in a console only run mode.

[IronBits]

Originally posted by PCZ
What about the via nics built in to the km266 chipset.
Do they net boot ? Looky here what I found...
List of PXE supported NIC/Chipsets :D
http://www.bootix.com/us/products/promtypes.shtml

[Paratima]

Just posted this on the "News & Discussion" thread also, seein' as how it's the current hot topic.

Check out this site (http://www.dnaresearch.com.au/gahnix.shtml), posted by dnar a while ago.

[IronBits]

Snagged this from an Ars thread -
Of course you don't wanna use the SETI client, but of particular interest was the addition of the nForce1/2 nVidia driver! :)

[This message was edited by Pjotr on July 30, 2003 at 18:24.]
Here is the Nero CD image file .NRG available on Knoppix + SETI CD image. (use passive ftp) If I knew how to make an ISO, I would have, sorry.

ftp://kometv.no-ip.com:4711/pub/SETI/NordisKnoppix-2003-06-27_seti.nrg

This is a Nordic Knoppix version 3.2 (defaults to English still) with 5 files added by me:

KNOPPIX\configs.tbz
KNOPPIX\knoppix.sh
KNOPPIX\nvnet.o (nForce1/2 nVidia NIC driver missing in Knoppix)
KNOPPIX\setiathome (i686-linux-gnu binary)
KNOPPIX\user_info.sah

These are in the KNOPPIX dir on the CD. Basically the only difference from the default Knoppix CD is the installation of the nVidia NIC to detect nForce NIC correctly and the setiathome client starting at boot with nice -19 priority.

What you must do to adjust the CD for your SETI account is to unpack the CD to your HD. Replace the user_info.sh and optionally change the knoppix.sh script to use another SQ proxy than mine. IMPORTANT: Do not edit the .SH file in Windows Notepad or similar, you must NOT create a Windows CR/LF text file, it must stay LF only! To be safe, leave the .sh file to use my SQ, just replace the user_info.sah file.

After you have done this, you burn a new CD. Create a Bootable Data CD, select the file KNOPPIX/boot.img as a 1.44 floppy boot image in your burning software. Burn the CD and you're done! Insert into any working PC and reboot it, it will start to crunch SETI for you until you power off. It will lose any half processed WU when you power off as it's running from RAM disk.

[DocWardo]

I admit I didn't read the whole thread here. but just wanted to share my 2 cents on nforce nic's.

I recently took a msi nforce 1 board and using the nvidia drivers. it sucked. that's all I can say. the drivers used up so much of the CPU that gnome2 (yeah this is my desktop box) was all jumpy. Ended up using a dlink PCI and it worked like a charm.
I have heard similar problems with the nforce 2 series.

maybe it was just my setup. or just that MSI board. but it was darn frustrating till I figured it out.

[rsbriggs]

Please let me know if someone comes up with an *EASY* way to get things to network boot, OR even just a WAY to do it for that matter.

(What software is required on the server, and how do you build the boot image that is going to be downloaded to the clients? Need to install TFTP? DHCPD? PXE something or another? LTSP? )

I wouldn't even mind having a CD drive on my network boot boxes, I suppose, which would be great if I could find or work out a procedure to:

0.) Start with a live CD, i.e. the Knoppix CD. somehow transfer to HD

1.) add any NIC driver that might be needed
2.) add in the DF software, and the correct handle file
3.) add in an autostart DF script
4.) re-master the new, bootable, CD....

Then duplicate the CD, place copy in various machines and boot - crunches DF until CD removed and re-booted.

[Paratima]

MR. BOB!

I'll say it again. Complete instructions & code are HERE (http://www.dnaresearch.com.au/gahnix.shtml) . :cool:

[PCZ]

Paratima
We are trying to network boot. The link you posted describes how to boot from a floppy. Not the same thing.

[magnav0x]

It's the same thing PCZ. It's just like netboot with PXE but you use the software on the boot disk instead of PXE (better for those with nodes that don't have PXE).

http://ws9.jobnegotiator.com/html/netboot6.html

Read up on that page and see if you feel up to doing diskless boots.

[PCZ]

If a floppy is used to boot up the node then it is not a discless workstation.
I do not want to use floppies.

[Paratima]

Why sweat PXE if ya don't have to? :D

Dummies like me gotta have it simple. That FDD spins for about a minute, you contact the Mother Ship & Bingo! you're loaded & crunching!

[rsbriggs]

Originally posted by Paratima
MR. BOB!

I'll say it again. Complete instructions & code are HERE (http://www.dnaresearch.com.au/gahnix.shtml) . :cool:

Bookmarked!! Thanks for the link It appears that diskless booting and floppy booting are very similar. In the diskless case, you use PXE or similar to download the boot image, via DHCP and TFTP (?). With floppy boot - you have the boot image on the disk and just load it directly. One could be converted to the other fairly easily, and you can consider floppy booting as being the step just before getting network booting working. Once you have a floppy image that boots - next step - send it over the network.

I'm going to go spend a few hours reading now.....

[rsbriggs]

Originally posted by Paratima
There's more? :confused: :jester:

Yeah - you forgot cp and man :D

[Paratima]

:rotfl: :rotfl: :rotfl:

[rsbriggs]

I'm not too certain that I understand why Apache has to be set up. Is the boot floppy somehow using http to download the files after it boots? While the link is interesting, if for no other reason than the explanation of how to NFS mount things, it raises more than a few questions in my mind....

[Paratima]

Yes, Apache is the HTTP server for the program loading. And I just heard that I have an emergency at work. Try emailing dnar direct. He's listed in the member directory, although he hasn't posted in a while.

I'll be back................... :looney:

[CodeMonkey]

Well, when you get back, (assuming no one else knows the answers) answer me a couple of questions about Gentoo.

Spent the last two days (again!) building a system for my AMD box. Booted it up and find that I don't have any network. I guess that means I need to load a module, but I DON"T KNOW WHICH ONE.

Now, the Gentoo Live CD boots up, finds the appropriate module for the NIC and loads it, so I know it exists and works with this Biostar MOBO.

Q.) How can I look at the Live Gentoo system running from the CD, and find out what NIC module it loaded? I'm assuming that I can just modprobe it on the system I built to see if A. I have it, and B. that it works. If it does work, I can just add it to be autoloaded, and I'm in business....

Q.) I forgot the other question.... Probably something along the lines of, can I just emerge Apache and end up with it all set up and running? Or, as I remember, is there a great deal of configuration that needs to be done first?

Sorry - I'm getting punchy - about 4 days in a row now with very little sleep trying to get this thing gen'ed....

[Dyyryath]

You can boot the Live CD and type 'lsmod' at the command prompt (as root) to see which modules are currently loaded.

Of course, you can also do 'modprobe -t net' to have modprobe attempt to load every module in the 'net' directory (your network drivers).

As for apache, you pretty much just need 'emerge apache'. The document structure will go in /home/httpd and the configuration structure will be in /etc/apache2. The default install will get you up & running. You can start it by doing '/etc/init.d/apache2 start'. Then check http://localhost in your browser to make sure it's working. If you want it to start on boot, then do ' rc-update add apache2 default'.

[magnav0x]

If you can't figure it out, just post what kind of network card you have and we can probably tell ya.

[rsbriggs]

Well.... lsmod shows "nvnet" as the network module. Of course, I can't seem to find anything in the kernel config that applies. It's an Nvida-on-board card....

Tried just moving it from the Live CD to my system, but there are a batch of unresolved symbols when I attempt to load it.

[magnav0x]

You need to get the nvidia drivers installed:

emerge nforce-net
cd /usr/src/linux
make menuconfig

Then complie the kernel with realtek support:

Network device support --->
Ethernet (10 or 100Mbit) --->
## make sure that EISA, VLB, PCI and on board controllers is
## checked, go down to this
<M> Realtek RTL-8139 PCI Fast Ethernet Adapter support
## press 'M' for it to be compiled as a module

Save the kernel configuration and recomplie the kernel:

make dep && make clean bzImage modules modules_install

Then mount your /boot partion if you need.
cp /usr/src/linux/arch/i386/boot/bzImage /boot

Then modify modules.autoload file to load the module on boot:

echo 8139too >> /etc/modules.autoload

Reboot and pray that it works :harhar:

You may or may not need to emerge nforce-net, but I would ASSUME you would.

[rsbriggs]

Realtek is compiled in. I even tried to modprobe the 8139 drivers - no luck. Well, I'm going to give this one last try. Let's see - no network, so have to:

boot from the CD,
mount stuff,
chroot,
emerge the nv-net stuff and
kernel build I guess.....

Sigh....

[magnav0x]

You may be able to just emerge the drivers and do it without recompiling the kernel.

Try:

emerge nforce-net
echo nvnet >> /etc/modules.autoload
Then recomplie the kernel.



Reboot and see if that works. If not then resort to my previous method.

[rsbriggs]

emerge nforce-net
!!! all ebuilds that could satisfy "nforce-net" have been masked !!!!
please correct the problem and try again....

having to boot from the Live-cd - remember I lost network connectivity....

[magnav0x]

Third option (if you don't want to emerge...dunno why you wouldn't) is to get the drivers straight from nvidia and transfer via floppy or cdrom.

http://www.nvidia.com/object/linux_nforce_1.0-0261.html

Download
Kernel Tar File NVIDIA_nforce-1.0-0261.tar.gz

And follow the directions.

[magnav0x]

Ok to get masked emerge packages you need to add a file to your system.

cd /etc
mkdir portage
cd portage
echo "net-misc/nforce-net" >> package.unmask
emerge net-misc/nforce-net

[magnav0x]

BTW if it still gives you the error then:

ACCEPT_KEYWORDS="~x86" emerge net-misc/nforce-net

[rsbriggs]

The good news, did an emerge sync, followed by emerge nforce-net, and all is well as far as network connectivity goes !!!

Not going to try to do much more this evening - too frazzled.... Is there a way to update everything? like emerge -up world ???

Thanks for the help....

[magnav0x]

This would be the quickest route:
emerge -u world

but I would suggest running

emerge -UD world

U will upgrade only...so it won't accidently downgrade a package and D will do a deep search on dependencies rather than just the obvious dependencies.

Glad you were able to get your network card up and going. Just remember emerge world is gonna take a long time so let her go and get some shut eye :cool:

[rsbriggs]

Didn't find much to update - only a couple of files...

Umm, let's see...It's just doing GCC headers and the kernel. :eek:
Guess I'll see how it's progressing in the morning.. ;)

[PCZ]

Progress

Got the first node to network boot :bouncy:

[rsbriggs]

Excellent! That means that you have the server set up, right?
Running what on it? DHCPD and Tftp ??

[IronBits]

Originally posted by PCZ
Progress

Got the first node to network boot Do tell do tell!!! :)
w00t!
Are you using linux or windows based server running dhcp/tftp ?
:cheers:

[rsbriggs]

Yes. Yes. Details, please. Looks like I'll be starting a Gentoo build on one of my big server boxes this evening, soon as PCZ releases his magic receipe......

(Or if it's Winders, then I'm already ahead of the game by a day or so....)

It's starting to feel like double the boards running soon..... :D :hifi: I'm already having visions of my name in orange letters in the stats..... :thumbs:

IB - what board was it you keep crowing about ??? I'm already getting ready to place a Newegg order. Oh wait - better find out from PCZ what board it was he just brought up, exactly....

[PCZ]

The first board up is an Asrock K7VM2 It uses the same VIA chipset as the boards you have ordered IB.

The Server is Red Hat 9 and I am using the software from LTSP.org.

I am going to try and boot an Nforce 2 board now.


PS

I don't need no stinking floppy!!

[rsbriggs]

Ok - so let me get the process down so I can get started this evening. Install Redhat (or Mandrake - it seems LTSP supports either).

Grab the LTSP .rpm from where? Which? Core? Kernel? Both? More than both?

Config files that need to be modified?

How are you getting the DF app onto the board, and starting it, or are you doing that yet?

[PCZ]

rsbriggs

Read This (http://www.extremeoverclocking.com/articles/howto/FAH_Diskless_Farm_1.html)

It works great as long as your nic is supported.
I used it with Intel and Via nics and it works great.
The nvidia nics are a different story however and I will require the assistance of a guru to make a new kernel.

[rsbriggs]

New kernel is not that much of a problem... Finding how to get nforce-net added might be another issue. Although - the nVidia drivers are available on their website. I'm assuming that the kernel that gets downloaded is the one that has to be re-compiled.

Let me work through the link myself, I'll know by the time I get to the end, since I have an nVidia system to test with - should catch up to you in a day or two....

I'm going to work through this without using a separate domain or farm server, for the time being. I'm so tired of shuffling hardware around, that I'll wait until I actually have something working this time :)

[^7_of_9]

Onboard NIC's usually suck for Net booting for some reason ........ My K7S5A does too :nod:

[PCZ]

I have one small problem remaining.
The remote nodes mount a share from the server.
This contains the DF program. This is working.
The problem is it is read only.

I have changed permissions on the folder to give everyone read write and exectute but it still mounts as a read only filesystem.

[rsbriggs]

Can't answer that one off the top of my head - not certain why NFS mounts would always end up "ro". Check the file /etc/fstab and etc/mtab (?) on the client after you boot. It should be something that gets set on the server side - double check the /etc/exports file on the server, too... Seems like rw versus ro would be controlled there.

Hopefully Dyy, or one of the other Linux gurus will jump in here....

[PCZ]

I sussed it out.
I had an extra space in one of the lines in etc\export.

[Dyyryath]

I just got home and thought I'd pop in to check out how things are going. Damn look at you two go! :thumbs:

Before you know it I'll be asking you guys for help with Linux! :notworthy :D :D

[magnav0x]

Get some sleep guys, save some fun for later :fart:

[rsbriggs]

Way too early for bed ;)

Starting Red Hat EVERYTHING install on my server box...
Anyone know how to get RH 9.x to recognize my darn USB device, and automount it?

Mandrake does, but it doesn't recognize the on card NIC on this board.
RH recognizes the NIC, but I can't get it to automount my USB devices....

Sigh... Can't they just put all the right stuff together in one distro ??????

[rsbriggs]

Looks like it has to be Mandrake 9.1 - my Red Hat install disk #1 seems to have gone bad.

[rsbriggs]

(another very large sigh)

Unfortunately, I'm really striking out with Mandrake.

Following the instructions, and attempting to install bind gives me:

1.) a conflict with the tmdns package - bind can't be installed.

2.) There is no dhcp .rpm file available for Mandrake, only sources, and the source RPM doesn't appear to build under Mandrake.

AAAAAAAAAARRRRRRRRRRRRRGGGGGGGGGGGGGHHHHHHHHHHHHHH.

Let's see - Red Hat installs have gone bad.

Gentoo has ltsp-core available as an emerge, but not ltsp-kernel, and it's an old copy of ltsp core that brings in all of X, and a zillion other things, including sound card modules....

Looks like I just can't win tonight. Guess I'll try burning another copy of the first Redh Hat install disk tomorrow and try again.

[Paratima]

Hey, Mystery Man, I'm back. Well, more or less. :p

Try THIS (http://www.suse.com/us/private/download/suse_linux/index.html) ! Most complete distro around.

[IronBits]

Hell yes! Linux got the highest rating possible! :smoking:

Linux, running on IBM computers using Intel Corp.'s (Nasdaq:INTC - news) chips, received the Common Criteria certification, a global standard for security features and capabilities of information technology products, IBM and SuSE said.

This means is that government can consider Linux when making purchasing decisions. :thumbs:
http://story.news.yahoo.com/news?tmpl=story&cid=569&ncid=738&e=1&u=/nm/20030805/tc_nm/tech_ibm_linux_dc

[rsbriggs]

Originally posted by Paratima
Hey, Mystery Man, I'm back. Well, more or less. :p

Try THIS (http://www.suse.com/us/private/download/suse_linux/index.html) ! Most complete distro around.

Sigh.... Maybe tomorrow...... Do you run SUSE?

[CodeMonkey]

http://dag.wieers.com/home-made/apt/

FWIW:

This is a link that has "apt" for Red hat, and a repository of packages that are apt-get ready and available for Red Hat 7, 8, and 9

:)

[Paratima]

Originally posted by rsbriggs
Sigh.... Maybe tomorrow...... Do you run SUSE? I admit to being a SuSEr, as is (or was) Chinasaur. I've had very few problems from it, although they do keep some system files in different places from Red Hat. (I keep a cheat-sheet handy.) ;)

My company uses Red Hat servers at customer sites around the country, which I set up and sometimes admin, but the box on MY desk runs SuSE 7.3. The other box on my desk runs Win2K, for corporate compatibility. At home, it's the same combo, W2K & various generations of SuSE.

Oh, and thanks for that last link, CodeMonkey.

[PCZ]

ALL
I have a problem with the size of the ramdrive.
I can only make it a maximum of 4mb.
/tmp is mounted on the ramdrive.
As DF needs about 10mb of temp space I have had to mount /tmp on an NFS share on the server.
This works and DF starts up and runs automaticaly.

Will I need to recompile the kernel to support a larger ramdrive ?

[rsbriggs]

And I'm so discouraged that I'm about to just shut off all my computers. Period.

I can't get ANY of this crap to work together. It's either the distros don't work with ltsp, or the PXE doesn't work, or something. I've now tried floppy booting, pen drive booting, network booting - can't get s**t to work....

PCZ - tell me, when you downloaded the ltsp stuff, did you download the exact versions mentioned on the web site, or the newer versions? What did you install *EXACTLY* ? If RedHat - what TYPE of install - workstation, server, minimal, everything, what ???

Can anyone tell me if DF works well in an openMosix environment? Is an openMosix cluster a possibility? Theoretically, it supports network and PXE booting. Thinking about giving something like clusterKnoppix a try, before I just turn all these boxes off and just bag it.....

[PCZ]

rsbriggs

I downloaded the files from LTSP (http://ltsp.org)

ltsp_core-3.0.9-0.i386.rpm
ltsp_kernel-3.0.9-0.i386.rpm
ltsp_local_apps-3.0.0-0.i386.rpm
pxestuff-3.0.5-i386.tgz

The rpm's are installed like this:
rpm -ivh ltsp_core-3.0.9-0.i386.rpm

The pxestuff archive has a readme in it telling you where to put the files.

I installed Red Hat 9 with the custom option and selected everything.
I didn't really need all the apps but I was playing safe.

I edited the config files using Jasons article as a guide.
I can help with the config files if you get stuck.

[Dyyryath]

Don't give up yet, rsbriggs! If things get ugly enough, I'll set one up here at the house (I've had some experience with net booting in the past) and then detail *exactly* how to get it working with a specific distro.

You guys seem like you're making good progress, so I've been content to keep chugging away at my benchmarking stuff (man is it slow going) and wait to see what you come up with. I've got faith in you both, but if you get well and truly stuck, I'll certainly jump in with you. :thumbs:

[IronBits]

Well some one get 'something' to work!
I struck out on the winblows stuff, unless I want to drop $3000, NOT!
On the first M7VIQ mobo, the nic is not recognized in w2k :cry: and I can't find a driver that works, yet...
I have 4 more to throw together tonight, and I'll roll that 'image' drive to the other 4 to see if any of the others will work, just to make sure it's not driver related. I already went thru different known working cables and switch ports just to make sure everything else was on the up and up.
I'm glad I have the HDDs for back up, so I can at least get them crunching, off line if necessary, until the new 3Com NICs get here ;)

Keep digging and clawing and fighting to get this working - it's not just for yourself.
If someone can get a write up on how to roll your own setup, others would be encouraged to give it ago, and more nodes would be brought online :thumbs:
We need more nodes, more members, more instructions :D

:cheers: Go Team Free-DC Go :cheers:

[PCZ]

IB

I have it working.
If you want to go ahead with it on your new MB's then lets do it.
I only have a problem with nvidia nics.

[rsbriggs]

PCZ - what boards are you using to connect with ??

My current "progress" - my M7NCD-Pro board hangs during PXE boot - says "options not recognized". My VIA cards actually manage to netboot, but won't run the DF app. I think I have things more or less set up right now. I installed the K12ltsp distro, which is RedHat with ltsp added in.

Now what I need to find is a board that will both netboot AND run the DF app, and buy a crapload of them....

It is possible that I can somehow update the flash on this Biostat board and get it to work. Right now, I'm so frustrated by the whole process that I am quite willing to spend some cash at NewEgg for boards that DO work WITH the on-board net card.

[PCZ]

IB
what nics are they ?

Do they say VIA rhine on bootup
Enable boot from Lan and see if you get the PXE boot loader.

[PCZ]

rsbriggs

The reason DF won't run is because /tmp is too small.
DF needs about 10 meg.

Do a df -k and see how big the /tmp mount is.
I guess it is only 1 meg.

This is easy to fix by mounting /tmp on an NFS share rather than the ramdrive

I can give you the config to do this if you want ?

[rsbriggs]

No, the problem on the VIA EPIA cards is "illegal instruction". The DF application is compiled in such a way that these mini VIA cards can't execute it. It's a known problem.

The problem with my Biostar board is that the PXE stuff is NOT right - they just hang during loading, and I don't know how to fix it. The machine doesn't have a floppy, so I can't upgrade the BIOS....

Please tell me what motherboard you got to successfully connect and run. I DON'T have the /tmp too small problem as far as I can tell - with what I'm using, it's a variable sized ram disk that grows as necessary. I'll gladly go out and buy another MOBO, just to see if I can get this to work. Once it does, I'll go out and buy 1/2 dozen more....

[PCZ]

OK
I am using ASrock K7VM2 MB's
They have a VIA rhine nic built in and they work fine.
The chipsets are via KM266.

How did you set up the variable size ramdrive I can only go to 4 megs. ?

[IronBits]

Well, the NIC does not work with w2k on the 2nd board :cry:
I fired off a message to their tech support to see if there is something I'm missing.
I can ping 127.0.0.1, but there is no RX/TX going on.
It is possibly related to the onboard video and it's shared memory thing.
That will be my next step to test while I wait for thier tech support to respond with other suggestions.
Biostar M7VIQ - yes, there is an option to make it boot from NIC using PXE 2+ ;)
Until I find out what is stepping on the NIC I/O - I can't do much else...
If I tell it to boot from the NIC, I see Rhine II 2.13 :D

[PCZ]

IB
Thats the same nic as my ASrocks.
I had trouble with the nic in 2k until I loaded this driver (http://www.asrock.com.tw/Drivers/VIA%20Chipset/LAN/LAN.zip)

You should be set for the Linux netboot.

[IronBits]

I'll give that a try! All mobos failed on NIC under w2k. Shows 100mb link, but no traffic :( :bang:
brb and let ya know how it went :)

[rsbriggs]

OK - I'm at the point where I can actually run something booted off the network. However, I can't ping anything locally from the network booted comp, except the server at 192.168.2.254.

How do I get the farm server to route packets from the 192.168.2 subnet over to the 192.168.0 subnet ?? The link that the guide gives for ipmagq.cjb.net doesn't work anymore.....

[PCZ]

I can help with the IP setup
I need to know what your set up is there.

Do you have a router to the internet ?
Does your server have 1 or 2 nics.
What network does the rest of you PC's live on.
192.168.0 or 192.168.2 ?

[rsbriggs]

Windows network has shareable internet connection...

The dual NIC farm server that I just set up :

static IP on windows network as 192.168.0.2 (it can see the network OK)
static IP on farm network as 192.168.2.254 -> farm hub

nothing on farm hub can see network. I'm assuming they want their gateway to be set to 192.168.2.254, have to get it to route over to the other NIC somehow

Anyone need some cheap hubs?

I've got a whole batch (like 20) of 24 port 10-base-T Cisco SuperStack II PS-40 hubs I'll sell cheap. They are not fast (10 Mbs), but have lots of ports, and come in a nice 1-U case....

[PCZ]

Do you want to have 2 networks ?

It would be easier if everything was on the same subnet.
If you want to keep the 2 networks the server will have to be setup for routing.
It will also have to do NAT or your internet router wont let packets from your netboot nodes out.

[rsbriggs]

Well, the instructions showed how to set it up this way. But I can't do the IP masquerading part....

Trouble is, I HAVE to have the windows box at 192.168.0.1, and it handles DHCP requests to the comps on the windows network (about 6). I don't know how to have the farm server be on the same subnet, and not give out possibly conflicting IP addresses. I suppose that I could move everything over to the farm network, but right now this is a problem.

With two NICs, I still don't know how to get packets routed from one segment of the network to another. Even if I put two static IPs on the farm server, xxx.0.2 on the windows network, and xxx.0.3 on the farm network - there are still troubles getting packets from the farm side to the windows side....


Internet <> server1 <HUB-1>

<HUB-1> < 192.168.0.2>farm server< 192.168.2.254> <HUB-2> -> to the farm

[PCZ]

try this (http://ipmasq2.cjb.net/)

[^7_of_9]

Originally posted by rsbriggs
Well, the instructions showed how to set it up this way. But I can't do the IP masquerading part....

Trouble is, I HAVE to have the windows box at 192.168.0.1, and it handles DHCP requests to the comps on the windows network (about 6). I don't know how to have the farm server be on the same subnet, and not give out possibly conflicting IP addresses. I suppose that I could move everything over to the farm network, but right now this is a problem.

With two NICs, I still don't know how to get packets routed from one segment of the network to another. Even if I put two static IPs on the farm server, xxx.0.2 on the windows network, and xxx.0.3 on the farm network - there are still troubles getting packets from the farm side to the windows side....


Internet <> server1 <HUB-1>

<HUB-1> < 192.168.0.2>farm server< 192.168.2.254> <HUB-2> -> to the farm

What DHCP software are you using for Windows? Is it Windows 2000 Server? In that case just create a DHCP Scope to work around this Create an address pool up to a certain IP range and then do the same on the other end.


What kind of probs are you having with merging the networks?

We'll call the "Merger" system Machine one. This is the one that needs to help join the network.

Need two NIC's in it. Give them both Static IP addresses. One for one Sub and one for the other sub. That will be your starting point.

Then the best thing to do (As you may not get a proper answer from anyone fast enough when you like it is visit these next two URLs.)

http://www.learntosubnet.com/
http://www.learntcpip.com/

Great sites that helped me work along with the MCSE course.

[IronBits]

No joy on the networking with these BIOSTAR M7VIQ mobos
VT6103 lan chipset is what they have. Guess I'll have to wait until tech support answers my email :( :bang: :trash:

[PCZ]

That sucks

:bang:

[rsbriggs]

Thanks for the info everyone. I visited the IPMasquerade site, found a config file, and ran it on the server. The farm comp can now see the network.

My little VIA EPIA 800 Mhz board is now crunching (or attempting to, anyway) a seventeenorbust (SB) unit. (remember, it won't run DF...)

Guess I'll see how it goes in the morning....

BTW IB, My M7CND-Pro Biostar board will NOT network boot. There is something wrong with these boards or this brand of board, and/or the drivers on them.... :(

Good night all - had lots-and-plenty of excitement for the day, so much so that I "forgot" to go to work today.....

[IronBits]

Quickly picking up were we were left off ;)

PCZ - I would like to have the tftp server on a windows box if possible.

Once you and rsbriggs get me the 'how-to' so I can figure it out, I'll start working on it and document the hell out of it so anyone can do it ;)

I'd like to see how it's done on both *nix and Windows, I can do *nix here, but *need* to do Windows at another location if at all possible.

For now, I got the stack working with Win98. I did get Biostar tech support today, and they say it's the first they heard of it. He took my name and phone number and is going to set it up, just like I have/want it, and see if he can get the correct drivers :)

I told him that under w2k, it works great, sees the 100mb link, but then there is no TX/RX - kinda makes it hard to connect to things that way, ya know? :rotfl:

[rsbriggs]

Well, just put in order for K7-VM2 and AMD 2400+ at New Egg....

Hopefully the first of several. I'll just have to see how network booting this puppy goes.... Sheesh - when are the 2400+ chips going to come down in price some? :eek:

Also, going to try to set this up in an openMosix cluster. Start up 10 copies of DF on the server, and let the cluster management move them around to the various machines as needed to maximize performance. ( /me keeping fingers crossed that this will work - it would be SO cool....) Assuming that the K7 board just comes up, I will add more of them to my cluster until I run out of household power in my computer room and fuses start blowing. I'm already thinking about adding a window A/C unit to the room to augment the central A/C, which can't keep it cooled down. Hoping for winter soon.... :gone:

[PCZ]

ALL
I am still trying to get the Nforce 2 nic's to work.
I can get them to boot recieve an IP address and load the ramdrive
using pxeclient.o. {I upgraded pxeclient.o to get them to boot}
But when the kernel is loaded it fails because it can't find a driver for the nic.

I need to find a way to append a net module to the kernel or build a new kernel
with nvnet support built in.

Help !!! I'm drowing here

[rsbriggs]

Well, at the very least, you have to have the lstp kernel module downloaded and installed. Trouble is - I'm not sure how to preceed from there, since I'm at work. There must be a doc that describes how to re-compile the LTSP kernel once you have the sources installed.

Hey....

Don't they release TWO compiled kernels with LTSP - one that has what most people need, and one that they call the kitchen sink - that has about every option you can compile in??? Can you try using the kitchen sink version?


Also, they say:

If you are going to be building your own custom kernels for the workstations, then you will need the ltsp_initrd_kit package. This package contains the scripts and root filesystem hierarchy that makes up the initial ram disk (initrd) that is downloaded along with the kernel, when the workstation boots.

And another edit.

They just released version 4.0 of the ltsp package.

===bob briggs

[PCZ]

I have the init kit. Version 5 , 9 and 10
I don't think the kitchen sink version would help as it seems nobody puts nvnet support in.

It boot the initrd OK.

I just don't know how to make a kernel.
It needs to be a initrd kernel tagged for netboot as well just to complicate matters.
I downloaded the tar file from nvidia but I definitely need a life raft here.

[IronBits]

Well I am using win98se for the time being.
I heard back from the Biostar tech support, they did a fresh install (told me what drivers they used) I did all if it, with PnP and without PnP OS, and I still ended up at the same spot, so after 8 'fresh' installs, I gave up.

[rsbriggs]

OK - first off, tell me how you upgraded the pxeclient. My Biostar Nforce-2 board tries to start net-booting, but always hangs. I don't see any sign that it makes it clear through trying to load the kernel. How are you telling? All I can see is that it hangs after trying to figure out what the entry point is.

If I can get far along so that the only problem holding me back is that the kernel doesn't have nv-net support, I have no problem spending the weekend figuring out how to rebuild the ltsp kernel with it turned on (and it IS a kernel option that can be turned on. I know, because I managed to get my Gentoo kernel built so that it would use the NIC.. )

[PCZ]

Yes the entry point was where I was getting stuck before I replaced pxeclient.0.
Download this Archive (ftp://ftp.kernel.org/pub/linux/utils/boot/syslinux/syslinux-2.05.tar.gz)

Just take the pxeclient.o out of the archive and overwrite the one in /tftpboot/lts

[rsbriggs]

OK - I head for home in a couple of hours, can do it then.

I seem to recall that the K12 LTSP package uses pxelinux.bin or something - I'll get that figured out when I get home - don't know if that is a softlink to something, or what.

Assuming that I can get the kernel to actually start loading, it shouldn't be too difficult to figure out how to get the nv-net stuff compiled into the kernel that it loads. I did read something about netboot-tagging it....

I'm going to start keeping really good notes on the procedures as I go along. I always forget how to do things....

[rsbriggs]

You've probably read this, right ?

http://www.ltsp.org/documentation/pxe.howto.html

[PCZ]

I would really like to nail this nvidia thing.
I feel that it is doable.

IB

This explains how to do the windows TFTP
Read ME!! (http://diet-pc.sourceforge.net/windows/etherboot-w2k.html)

[IronBits]

Good find! Bookmarked ;)

[PCZ]

rsbriggs
The netboot thread is a good idea.

[rsbriggs]

I'm hoping to pull all the download locations and procedures into a single place, so that it can be written up at some point....

(And I'm getting really tired of forgetting where some piece of documentation is, where to download something, what thread I saw someone mention something or another, etc.)

Hopefully everyone will benefit this way :) :thumbs:

Feel free to add your message about what version of LTSP you used and the link to the download site for the files......

[IronBits]

Ok, I made your thread sticky - and I edited your first post, or rather, appended to it in hopes of keeping it an information resource.
This thread can be used for on going tech support, questions, tips and ideas :)

[PCZ]

I just started up 2 netboot nodes with fresh installs of DF and I am getting

MAIN SERVER IS DOWN FOR PROTEIN CHANGEOVER

Please try again later Press Enter

?

[rsbriggs]

Might mean that your internet access isn't working.

BTW: The NVNET driver only patches into kernel version .4.20 or later.

Now, the k12LTSP install uses the .20 kernel, which means that I could install that, and patch in the driver, and rebuild, which would get you the files you need - I think.....

Anyway, I'm thinking I'll just blow this install away, install the K12 stuff, and at the very least, move the 4.20 kernel, initrd and vmlinuz files out to my website so they would be available.

Since I'm seriously thinking about installing clusterKnoppix on Monday, this wouldn't be too big a hassle....

We need to get a patched kernel version 4.20 with nvnet.o installed in /opt/ltsp/i386/lib/modules/-kernel version-/kernel/drivers/net

Want me to give it a try?

[IronBits]

Originally posted by PCZ
I just started up 2 netboot nodes with fresh installs of DF and I am getting

MAIN SERVER IS DOWN FOR PROTEIN CHANGEOVER

Please try again later Press Enter ?
Can you ping www.distributedfolding.org? I hope it's not down for change over!

[rsbriggs]

Doesn't appear to be down. Think he just has internet access troubles from the netbooted clients (?)

[PCZ]

I am using a 2.4.21 kernel it came in the latest initrd kit.
It's not patched for nvidia though.

The nodes have Internet access. The message is coming from the DF servers.
I am going to try and start up a fresh windows install and see if the same thing happens.

[IronBits]

I just checked my clients and see no error messages about not being able to reach the Server...
I think it uses FTP ... port 20/21 stuff ;)
Check your PM.

[rsbriggs]

Where did you get the initrd kit ? LTSP?

[PCZ]

They are working now.
I think there was a small server glitch.

I got the kit here (http://prdownloads.sourceforge.net/ltsp/ltsp_initrd_kit-3.0.10-i386.tgz?download)

[PCZ]

IB
I'm feeling the HEAT as well England is very HOT at the moment I had to reduce the clock speeds of all my Athlons because they keep stopping.

You want a 2k server to do the tftp and dhcp.
The nodes will be running the ltsp linux image.

The server will also have to have shares that the nodes can mount.
The kernel only supports NFS I think so that could be a problem.
It is doable I think.

I have a 2k server here and the pxeboot software from Intel.
I will give it a try and see what I can come up with.

[rsbriggs]

OK - here is the deal. Basically the nforce stuff is not supported, except in kernel 2.4.20 (i.e Mandrake) by using the nVidia patches, or in the up-and-coming version 2.4.22. Seems there is some dissention in the ranks about nVidia using GPL code to build non-open source drivers.

The GART patches were supposed to be included in 2.4.21, but missed that release due to the controversy. They are supposed to be shceduled for release in kernel 2.4.22

As of this moment, there are only two version of Linux that support the nforce chip sets - Mandrake 9.1 and Gentoo....

The majority of the Linux guys basically say "you are often on your own, if you buy a board with a non-supported chipset...."

For now, there is no easy way to run Linux on those cards, other than using Mandrake.RedHat releases are currently using the 2.4.19 kernel and don't have nforce support available. Next Red Hat release might *possibly* have support for it. Doesn't appear to be any LTSP support available...

Bottom line - you (and I) are probably stuck not net-booting the board until the next kernel release.

For now, I'm going to put it in a case and run Win-XP on it. I'll net boot the other MOBO I have on the way....

[PCZ]

rsbriggs
I got the feeling that we were on our own with this one.

IB
I have found a freeware NFS daemon for 2k.
Also a good freeware TFTP server.

[PCZ]

IB
I am installing 2k server at the moment and will attempt to set it up for PXE booting.
I have downloaded some utils from the web which I think will do the job.

rsbriggs
I think we should leave the nvidia PXE boot thing alone until there is more support.

[rsbriggs]

I'll agree with that - support for it will be in the next kernel, and there are LOTS of boards out there that work fine, without all the hassle. I'll just run it in a case for now, and worry about getting the K7VM-2 motherboard working when it gets here.

[IronBits]

Thanks PCZ ! :thumbs:

[PCZ]

IB

will the server have more than one partition. ?
I am installing with 2 partitions c and d

[IronBits]

I always have two partions my friend... I see we think a like. :thumbs:
C$ belongs to the OS, everything else goes on D$ :D
I have 30+ gigs on D$ ;)

[PCZ]

IB
I haven't managed to get it working on 2k yet.
It isn't as easy as it first appeared.
LTSP is really meant to be used with a nix server.
I will return to this if I can find an example of it being done.

Also
I have had to shut down some of the pharmage because of the heat.
The Athlons keep stopping.
It was getting to be a full time job just keeping them running.

[IronBits]

Take your time, no hurry on the PXE thing.
All my boxen here have HDDs and PSs. I want to reduce the HEAT.

I will be putting two boxen on one PS in a few weeks to see how that goes, then convert the pharm over if it works well.

It's 104F on my porch right now, which meand 98F comming in from the 'cool' side of the house - it's HOT HOT HOT - I understand completely. :(

/me hopes my pharmage doesn't go into meltdown :scared:

[PCZ]

IB

I'll get back to work on the 2k boot server when I get some fresh ideas.
I will probably find some info on the web and revisit it with renewed enthusiasm.


I just had my Red Hat server go down. .:swear:

The ATX power plug had started to melt where all the red 5v wires connect.
I have changed the power supply and am keeping my fingers crossed it doesn't happen again. The original tigers don't have the auxillary 12v plug so a lot of current
gets drawn from the 5v rails causing the ATX connector to heat up.
I have used a better quality PSU this time and hopefully the ATX plug has better quality contacts.

[PCZ]

What is the best way to add DF to the startup config on the server.
I am starting the 2 clients on the server manually.

The nodes autostart DF
Sometimes they dont start properly but I have added
rm foldtrajlite.lock to the startup script and that helped a lot.

[rsbriggs]

bwkaz posted a script in one of these threads. Can't find it at the moment.

As a quickie, I'd put a script in /etc/rc.d and add a line to rc.local to start it. The other script actually ran it as a daemon that you could start/stop/restart, I think.

[the-mk]

The df-starting-at-booting-script is in this thread (http://www.free-dc.org/forum/showthread.php?s=&threadid=3822)

I'll quote the most important part for you:



#!/bin/sh

dfdir=/path/to/distribfold

case $1 in
start)
echo -n "Starting the Distributed Folding client..."
cd $dfdir
./foldit </dev/tty11 >/dev/tty11 &

if [ $? -eq 0 ] ; then
echo " OK"
else
echo " FAILED"
fi

;;

stop)
echo -n "Stopping the Distributed Folding client..."
rm -f $dfdir/foldtrajlite.lock

while ps -C foldtrajlite >/dev/null 2>&1 ; do
sleep 1
done

echo " OK"

;;
esac

[rsbriggs]

This tells you how to compile the kernel for LTSP. If you want nforce2 support, it has to be compiled into kernel 2.4.20.....


http://www.ltsp.org/contrib/customkernel.html

[rsbriggs]

How to set up PXE booting from a M$FT box.....

http://pxes.sourceforge.net/howtos/ms_only_environment/

[PCZ]

I would like to be able to telnet the remote nodes.
You can telnet from them but not to them.

I am starting them up in runlevel 3 so I can run commands.
I dont want to have to use a screen and keyboard.

[rsbriggs]

xinetd starts the telnet server - check out /etc/xinetd.conf.

I can telnet to the clients, but can't log on - something else is wrong:

YPBINDPROC_DOMAIN: Domain not bound

Also, I keep getting NFS errors and timeouts..... Frankly, I'm not very encouraged by the behaviour of the whole thing.

[PCZ]

rsbriggs

NFS errors and timeouts.
I suffered from this until I worked out what was causing it.

When I set up the server and the first node they were plugged into a 10/100 HUB.
This worked fine.
When I rollded out some more remote nodes they were plugged into one of my Switches. This is when I started to notice the NFS timeouts.

After a lot of headscratching the penny dropped.

HUB's only support half duplex connections, Switches support full duplex connections.
The nic's were making full duplex connections when plugged into the Switch.
Plugging the nodes into a HUB instead of a Switch cured the NFS timeouts.

The drivers for the nic's don't appear to work properly with full duplex connections.
The nics need to be forced to half duplex. (I'm not sure how to pass the required parameters to the driver)

I took the easy way out and plugged them into a HUB

[PCZ]

IB
Ok what do you want to know?

I am glad you are going for the single network.
Starting with dhcp.conf would be a good idea.

You have my complete and undivided attention for the next couple of hours
ask away.

[rsbriggs]

Well, I'm not IB, but I DO have a question. Using a single sub-net, how can you be certain that the *LINUX* server is going to answer the DHCP requests, rather than, in my case, the Windows box acting as an internet connection sharing (DHPC) server....

[PCZ]

You cant be certain unless you turn off one of the DHCP servers.
What is your internet connection Cable, ADSL or Modem.

[rsbriggs]

Connection is to a MICROSOFT DSL. That means, I have a dedicated box to act as a router to share the internet connection, and to connect to the DSL modem:

[modem] 192.168.1.1 <-> 192.168.1.2 <---> [windows box] <---- rest of network

The windows "router" box has a static IP of 192.168.0.1 and is the gateway for the rest of the network.

I don't know if it will affect my internet connectivity or not, but I'm guessing that I could turn the DHCP service off at that box, and let the Linux server do DHCP for the network. So long as everyone uses the 192.168.0.1 as a gateway, it SHOULD work....

The "rest of the network" also lives in the "MSHOME.net" domain...

And how would you turn off DHCP on the Windows box, I wonder???

[PCZ]

The problem is turning the DHCP service off without killing ICS.
It certainly runs a cut down DHCP service.

Can you see DHCP running as a Service ?

[IronBits]

The whole PXE network is over on an SNMP capable switch, server included.
It should resolve to it first, as dbestern.net.
If not, I'll disable the Windows 2000 DHCP until I can get it figured out.
Not a problem.

[rsbriggs]

IB,

The fact that you are hanging at the point of setting up the NIC means one of several things.

1.) The NIC isn't supported.
2.) There is a configuration error in one of your files.


Please email me at:

rsbriggs AT mailblocks DOT com

Once we make contact, I'll have you send me a couple of files, and I can double check the settings...

[IronBits]

It finds the NIC just fine, downloads the *nix kernal and everything else...
configuration is the problem, no doubt... NFS mount points maybe.
Prolly need to check ALL the configuration files :rotfl:

[rsbriggs]

http://www.linuxbios.org/

Looks like some people are taking diskless booting one step further. They flash Linux right into the BIOS.......

[IronBits]

Thanks to rsbriggs help tonight in IRC, I was able to bring up a few nodes!!!:notworthy

Now my /var/log/messages is filling up with the following
Aug 20 22:29:26 ws002 ypbind[221]: broadcast: RPC: Timed out.
Aug 20 22:29:29 ws001 ypbind[221]: broadcast: RPC: Timed out.
Aug 20 22:30:40 ws002 ypbind[221]: broadcast: RPC: Timed out.
Aug 20 22:30:43 ws001 ypbind[221]: broadcast: RPC: Timed out.
Aug 20 22:31:54 ws002 ypbind[221]: broadcast: RPC: Timed out.
Aug 20 22:31:57 ws001 ypbind[221]: broadcast: RPC: Timed out.
Aug 20 22:33:08 ws002 ypbind[221]: broadcast: RPC: Timed out.

I don't want to setup an NIS server to, and according to what I read elsewhere, it says I would need ypbind must run on every machine which has NIS client processes;

The quick work around for this is to DISABLE run_local_apps ...
Question is, will this PXE boot stuff work with it disabled?

[rsbriggs]

If I were you....

I'd be real tempted to set NIS_DOMAIN = "dbestern.net" in the default section of /opt/ltsp/i386/etc/lts.conf

and to comment out the ypbind/ypserve stuff in /opt/ltsp/i386/etc/rc.local

[IronBits]

NIS_DOMAIN = "dbestern.com" added to default section of /opt/ltsp/i386/etc/lts.conf

Fixed /etc/hosts and commented out the ypbind/ypserve stuff in /opt/ltsp/i386/etc/rc.local.

No more errors about ypbind.

The Server, and all the workstations, can not contact the DF server to get work, or upload work...
nslookup works, as does getting around the internet with a browser.

Where is it that I build the /etc/hosts file for each workstation?
They have an OLD incorrect one when looking at it from the workstation side.
On the Server side /tmp/hosts looks fine
when I go into /opt/ltsp/i386/etc and look at it there, it's ln to /tmp/hosts and it looks correct, but the workstation is wrong still.

[rsbriggs]

I *THINK* that you can safely turn off local apps. Ya just need to make certain that /home gets mounted for the local execution scheme that I use, is all...

[IronBits]

Where is it that I build the /etc/hosts file for each workstation?

[PCZ]

/opt/ltsp/i386/etc/hosts

[IronBits]

That is a link
#file hosts: symbolic link to /tmp/hosts

[rsbriggs]

Actually, because of the way things are set up, you'd have to do whatever you want to do to the clients /etc/hosts file in /opt/ltsp/i386/etc/rc.local

look at the way it builds the file domainname..

I'd guess there is something like
cat ${IP} {$HOSTNAME} > /etc/hosts
in there somewhere....

[IronBits]

Originally posted by rsbriggs
xinetd starts the telnet server - check out /etc/xinetd.conf.

I can telnet to the clients, but can't log on - something else is wrong:

YPBINDPROC_DOMAIN: Domain not bound
Was there a solution to the YPBINDPROC_DOMAIN thing?

I have PXE boot working with SuSe 8.2 ! :D

[rsbriggs]

There IS a solution to not being able to log onto the clients. Here's the (simple) problem, since we aren't setting up NIS, there isn't any password file on the clients, so no-one can log on.

Just copy /etc/passwd and /etc/shadow so that clients will have them when you try to log on.

cp /etc/passwd /opt/ltsp/i386/etc
cp /etc/shadow /opt/ltsp/i386/etc

I just commented the ypbind and ypserv stuff out of the clients rc.local file. Don't care about NIS very much....

[IronBits]

Still wasn't able to get DF to run without using the -i f switch.
I also had to increase the RAM disk size to 4096 from 1024 so there was enough disk space for the TMP files it uses with this protein when using the -rt switch.
It's running now, albeit off net! :D
The Server or the client cannot connect to the DF server to fetch/upload work.

ERROR: [000.000] {foldtrajlite2.c, line 4616}
Error during upload: NO RESPONSE FROM SERVER - WILL TRY AGAIN LATER
ERROR: [000.000] {ncbi_socket.c, line 1216} [SOCK::s_Connect]
Failed connect() to www.distributedfolding.org:80 {errno=101,Network is unreachable}
ERROR: [000.000] {ncbi_connutil.c, line 799} [URL_Connect]
Socket connect to www.distributedfolding.org:80 failed: Unknown
ERROR: [000.000] {ncbi_socket.c, line 1216} [SOCK::s_Connect]
Failed connect() to www.distributedfolding.org:80 {errno=101,Network is unreachable}
ERROR: [000.000] {ncbi_connutil.c, line 799} [URL_Connect]
Socket connect to www.distributedfolding.org:80 failed: Unknown
ERROR: [000.000] {ncbi_socket.c, line 1216} [SOCK::s_Connect]
Failed connect() to www.distributedfolding.org:80 {errno=101,Network is unreachable}
ERROR: [000.000] {ncbi_connutil.c, line 799} [URL_Connect]
Socket connect to www.distributedfolding.org:80 failed: Unknown
ERROR: [000.000] {ncbi_http_connector.c, line 101}
[HTTP] Too many failed attempts, giving up
suggestions?

[rsbriggs]

NETWORK IS UNREACHABLE seems to be pretty telling here.
The ramdisk is probably still too small - PCZ recommends using 32768

what happens when you ping www.distributedfolding.org ?

[PCZ]

From the bash prompt on one of your nodes type:

traceroute www.distributedfolding.org

The IP address of the first hop should be your router.

[IronBits]

It does...

traceroute to www.distributedfolding.org (38.112.100.76), 30 hops max, 40 byte packets
1 192.168.2.1 0.790 ms 0.510 ms 0.530 ms
2 66.51.209.1 12.920 ms 15.789 ms 12.852 ms
3 g6-0.mon.cr1.lax1.dslextreme.com (66.51.199.133) 12.952 ms 12.823 ms 12.787 ms

[rsbriggs]

Funny thing is, he can ping it just fine. But the client just hangs on the "checking for new version".
Wish I knew what the handshake sequence was, I'd just write something in 'C' that could diagnose the problem.....

[IronBits]

It's NOT funny! :bang: :trash: ;) :)

[PCZ]

If you can traceroute to the host then the network is ok

2 things both important.

1:
Permissions
Make sure all the files are owned by root and the group is root.
You do not want any files owned by any other user .


2:
Swap file needs to be bigger.
At least 16meg

[rsbriggs]

Mine is similar - you can see it going to the IP Masquerade box, then the gateway, then the DSL modem, then out the door:



1 server (192.168.2.254) 0.712 ms 0.316 ms 0.484 ms
2 DELL-600SC.mshome.net (192.168.0.1) 0.647 ms 0.523 ms 0.434 ms
3 192.168.1.1 (192.168.1.1) 1.332 ms 1.186 ms 1.152 ms
4 63.230.10.254 (63.230.10.254) 46.915 ms 47.335 ms 59.747 ms

[rsbriggs]

Last I knew, he was using NFS swap, and setting ram disk size to 4 megs....

But main problem - he can' even connect to the DF server via his main Linux server. I think clients would be OK if he can get his server connecting....

[PCZ]

PS
I have Telnet working.

I can telnet into the clients and run programs.
The program I need to run the most doesn't work however.
shutdown doesn't do a dam thing. :bang:

[PCZ]

I use run level 3
local apps are off.
nfs swap is off.

[rsbriggs]

It's a non-functioning version that makes a "busy box" call. I tried to move the real shutdown command over, but it wants libraries, and I didn't want to follow the whole thing through once I hit glib.... DIdn't try reboot command, but that's probably just a "shortcut" to shutdown -r now. Let me check what libaries it wants ( ldd reboot )...

[IronBits]

I have asked Master Dyyryath for some help on this problem of having ONE NIC card, and all three OSs failing to work correctly.
RedHat Shrike -- SuSe 8.2 -- Mandrake 9.1

Scenario:
one NIC
IP: 192.168.2.20
Gateway: 192.168.2.1
DNS: 192.168.2.2
DHCP: 192.168.2.20
Kernel IP routing table


Destination Gateway Genmask Flags Metric Ref Use Iface
192.168.2.0 * 255.255.255.0 U 0 0 0 eth0
127.0.0.0 * 255.0.0.0 U 0 0 0 lo
default 192.168.2.1 0.0.0.0 UG 0 0 0 eth0All this works fine, browsing, nslookup, ping and even traceroute - all works fine.

DF client can not connect to the DF server to get work, or upload work.:bang: :trash:

[Dyyryath]

Actually, it can connect, but for some reason, it's not getting a proper reply to it's HTTP GET request. I'm not sure quite why just yet.

The way it's supposed to work is:



(1) Client connects with SYN to DF server to start a connection
(2) Server replies with SYN ACK
(3) Client replies with ACK to complete 3 way TCP handshake -- connection is open
(4) Client issues HTTP GET request for /server.status
(5) Server replies with current date: 'Date: Sun, 24'
(6) Client replies with ACK
(7) Server replies with FIN ACK to close connection
(8) Client replies with final ACK

(9) Client sends SYN to DF server to start another connection
(10) Server replies with SYN ACK
(11) Client replies with ACK to complete 3 way TCP handshake -- connection is open
(12) Client issues HTTP GET request for GET /cgi-bin/trajstore?handle=(HANDLE GOES HERE)
(13) Server replies with current date: 'Date: Sun, 24'
(14) Client replies with ACK
(15) Server replies with FIN ACK to close connection
(16) Client replies with final ACK


Unfortunately, IB's box isn't getting past #5. He's never receiving the proper response to his HTTP GET request. The server is simply closing the connection with FIN ACK. Thus, his client sits there confused and eventually says it can't connect to the server. The fact is, it *has* connected to the server, it's just not heard what it was waiting for, and I'm not sure why.

I've gotta run to the office for a bit, but I'll be back in an hour or two. When I get back I'll tinker with this some more and see if I can figure out what's going on...

[PCZ]

IB
Which internet router do you have ?

There might be a bug in the nat translation.
Strange that it works in windows though.

[rsbriggs]

Don't be so certain that it IS actually connecting....

Have you verified that the HTTP GET is actually making it out? I had a similar situation when I tried to run using one subnet. The Internet Connection Gateway box simply wouldn't send or route tcp/ip packets for IP addresses that it hadn't assigned.

Ping, no problem. UDP packets, no problem. ICMP packets, no problem.
TCP/IP packets - oops - roadblock - they never made it past the ICG box out to the DSL modem. It refused to pass them.

My solution, as you know, a dual-homed box with two NICS and IP Masquerading. Works fine. Never did figure out how to get the ICG to pass ANY tcp/ip traffic from machines that it didn't send DHCP set up info to. Some Windows security deal, I suppose....

Oh - but didn't you say that you COULD browse from the server???
In that case, I'm mystified, since it's sending lots of HTTP GETs and getting responses. :confused: :confused: Never Mind, I guess....

[Dyyryath]

The problem was in his router. Actually, it's an SMC switch if I remember correctly. At any rate, I set him up a multi-homed Linux box as a router/gateway/firewall (netfilter) and changed the default route on the internal box to use it instead of his switch. It connected immediately. :thumbs:

We'll be using that same box to handle his LTSP stuff, too. I haven't gotten time to install that stuff yet, but it's working just fine as a gateway now, though it looks like I'm going to have to tweak his netfilter ruleset a bit yet...

[IronBits]

So there is no confusion, I use a SMC 7008 ABR Cable/DSL Broadband Router 10/100 Mbps Switch ;)
And kudos to Dyyryath for giving me a hand making a SECURE firewall :D

Now if I can only get him to finish it, and complete the docs on what he did on his end, so I can build another one just like it, then I'll do the write it up ;) :)

/me thinks Dyyryath plays too many games :rolleyes: :p :D

[PCZ]

IB
The SMC 7008 has a firewall built in. Thats why it is called barracade
One of the advanced settings is probably wrong.
Also I checked and there are firmware updates available.
I had to update the firmware in my DSL routers a while ago
to get netmeeting to work. So the firmware can make a big difference.

[IronBits]

OK, Dyyryath is off having a fun 3-day week-end I guess, as I
haven't heard from him, so,
I am attempting to continue on my own. I'm sure he can fix anything I fubar :)

I'm getting the following errror in /var/log/messages
ether xinetd[5403]: libwrap refused connection to tftp from 192.168.2.206
ether xinetd[991]: libwrap refused connection to tftp from 192.168.2.206
ether xinetd[14529]: Deactivating service tftp due to excessive incoming connections. Restarting in 2 seconds.
ether xinetd[14529]: Activating service tftp
Clues? TFTP is not handing off to the client.

[IronBits]

Ok, fixed that by editing the hosts.allow (using 192.168.0 instead of 192.168.2)
Now I get this
PXE-T01: File not found
PXE-E3B: TFTP Error - File Not found
PXE-M0F: Exiting Intel Boot Agent
Boot failure: System Halted

[PCZ]

You may have the permissions wrong on the files in the lts directory or the files specified in /tftpboot/lts/default aren't in the lts directory.
I sent you some updated kernel files .21 did you put them in the lts directory ?.

[IronBits]

Ya, got that all in there...
Some .cfg file is not pointing to where the tftp boot files are. :bang:
plus I just noticed there is no 'named' service on the server, do I *NEED* it?
if so, what is the urpmi 'name' that I need to install it?

[IronBits]

a HA! (maybe)
/etc/inetd.conf shows


service tftp
{
disable = no
socket_type = dgram
protocol = udp tcp maybe?
wait = yes
user = root
server = /usr/sbin/in.tftpd
server_args = -s /var/lib/tftpboot
per_source = 11
cps = 100 2
flags = IPv4
}
Shouldn't that be

service tftp {
socket_type = dgram
wait = yes
user = root
log_on_success += USER
log_on_failure += USER
server = /usr/sbin/in.tftpd
server_args = -s /tftpboot
disable = no
}

[IronBits]

Almost there!!! Now getting permission denied!

tail /var/log/messages shows
ether rpc.mountd: refused mount request from ws006 for /opt/ltsp/i386 (/): no export entry

client console says
Mounting 192.168.2.206:/opt/ltsp/i386 on /mnt failed: Invalid argument

[IronBits]

Ok, fixed that by editing /etc/exports
/opt/ltsp/i386 192.168.2.0/255.255.255.0(ro,no_root_squash,sync)
/var/opt/ltsp/swapfiles 192.168.2.0/255.255.255.0(rw,no_root_squash,async)
was 6
Now I need to fix my nfsmount points, as ws006 did not mount /home/ws006 at all.
/etc/exports shows
/home/ws006 192.168.2.6/255.255.255.0(rw,no_root_squash,sync)
but I didn't get there, it leave me in / (which is /opt/ltsp/i386)

[PCZ]

my nodes boot up to /
If I want to see the df files I need to cd to /DF

I created an empty folder called DF in /opt/ltsp/i386/
you have to do this it is a sym link.

Created a folder called DF off the root of the server /DF
and created sub folders called ws001,ws002 etc underneath.
/DF/ws001, /DF/ws002 etc

I made an entry in exports for each ws00* folder.
The DF/ws00* folders are mounted as /DF on the node.
this is done in /opt/ltsp/i386/rc.local

dont forget to restart the NFS service when you make changes to etc/exports

You can call the nodes working directory anything you like I used DF because it is the project I am currently working on.

## LTS-begin ##

#
# The lines between the 'LTS-begin' and the 'LTS-end' were added
# on: Tue Aug 5 08:08:30 BST 2003 by the ltsp installation script.
# For more information, visit the ltsp homepage
# at http://www.ltsp.org
#

/opt/ltsp/i386 172.31.158.0/255.255.255.0(rw,no_root_squash,sync)
/var/opt/ltsp/swapfiles 172.31.158.0/255.255.255.0(rw,no_root_squash,async)

/DF/ws001 172.31.158.0/255.255.255.0(rw,no_root_squash,sync)
/DF/ws002 172.31.158.0/255.255.255.0(rw,no_root_squash,sync)
/DF/ws003 172.31.158.0/255.255.255.0(rw,no_root_squash,sync)
/DF/ws004 172.31.158.0/255.255.255.0(rw,no_root_squash,sync)
/DF/ws005 172.31.158.0/255.255.255.0(rw,no_root_squash,sync)
/DF/ws006 172.31.158.0/255.255.255.0(rw,no_root_squash,sync)
/DF/ws007 172.31.158.0/255.255.255.0(rw,no_root_squash,sync)
/DF/ws008 172.31.158.0/255.255.255.0(rw,no_root_squash,sync)
/DF/ws009 172.31.158.0/255.255.255.0(rw,no_root_squash,sync)
## LTS-end ##




A piece of rc.local which mounts /DF


################################################################################
#
# Mount filesystems
#

if [ "Y" = "Y" ]; then
pr_set 85 "Mounting additional filesystems"
echo "Mounting additional filesystems..."
mount -t nfs -o nolock ${NFS_SERVER}:/DF/${HOSTNAME} /DF
fi

pr_set 86 "Setting up loopback device"
echo "Setting up loopback device"
ifconfig lo 127.0.0.1 netmask 255.0.0.0 broadcast 127.255.255.255

[IronBits]

ok, got it running finally - gonna have to wait on Dyyryath to help me figure out why it can't make it thru the gateway ...
For now, it's running off net :)

[IronBits]

from /opt/ltsp/i386/etc/rc.local
# Add Default Route
pr_set 87 "Adding Default Route"
echo "adding default route"
route add -net 127.0.0.0 netmask 255.0.0.0 lo
route add 0.0.0.0 gw 192.168.2.6 eth0

Do the clients use the same gateway as the Server?
If so, that would explain why using .6 won't work, as the gateway for the server says

Destination Gateway Genmask Flags Metric Ref Use Iface
192.168.2.0 * 255.255.255.0 U 0 0 0 eth1
66.51.208.0 * 255.255.252.0 U 0 0 0 eth0
127.0.0.0 * 255.0.0.0 U 0 0 0 lo
default 66.51.209.1 0.0.0.0 UG 0 0 0 eth0

outbound WAN NIC IP for the Server is is 66.51.209.58.
So would the clients use 66.51.209.58 as the gateway rather than the LAN IP of .6?
On the Server
eth0 = 66.51.209.58
eth1 = 192.168.2.6 <--- NFS/TFTP server to for the clients

Would the clients be using 66.51.209.58 as the default gateway?

[PCZ]

Your nodes will use the private address of the server as their gateway.
Your server should do the nat translation allowing them onto the internet.

so the gateway for the nodes should be 192.168.2.6

[PCZ]

go onto one of the nodes and type route

I need to know what they think their default gateway is.
Your nearly there !

[IronBits]

On the node, route shows
destination gateway Genmask Flags Iface
default 192.168.2.6 255.255.255.255 UGH eth0
There is no reference to 127.0.0.1 lo - but I can ping it.

I can't ping 192.168.2.6 or any place on the WAN, but I can ping any LAN node.
I can get nslookup results no problem from my LAN DNS server.

[PCZ]

I assume that the server itself can reach the internet.
We need to find out why 192.168.2.6 won't respond to ping.
This is the interface that the nodes boot from so it works for dhcp and tftp requests.
Also the nodes use this interface to mount the NFS shares.

Assuming the above is correct then the problem will be your iptables config.
you will also need to enable forwarding.

any chance you could post your iptables config

/sbin/iptables
/usr/sbin/iptables

/etc/sysconfig/iptables

I will be in one of those places.

[IronBits]

Oh crap :bang: Dyyryath is gonna kill me!!!
There are no ipchains configured :cry:
I thought he had set it up so that it would forward the packets.
Guess I had better wait until I can reach him again.
I looked in /proc/net/ip* and all are ZERO byte files.

I have NO clue about ipchains - an AOLer knows more than I about ipchains and I can't find Dyyryath :(

[IronBits]

I can't locate Dyyryath, as usual, when I need him.

WAN IP 66.51.209.58 gateway 66.51.209.1
LAN IP192.168.2.6

How would I setup ipchains? Dyyryath can fix it, if/when he gets time, if he's not too pissed of at me yet for messing up the security, if I messed it up.

# iptables -L -v
Chain INPUT (policy ACCEPT 154K packets, 52M bytes)
pkts bytes target prot opt in out source destination

Chain FORWARD (policy ACCEPT 0 packets, 0 bytes)
pkts bytes target prot opt in out source destination

Chain OUTPUT (policy ACCEPT 156K packets, 28M bytes)
pkts bytes target prot opt in out source destination

[rsbriggs]

This should get you through until you can find Dyy. It's what I use for IP Masquerading - you may have to fix the input and output eth settings. Your mileage may vary...


[root@DELL400SC root]# cat rc.firewall
#!/bin/sh
#
# rc.firewall-2.4
FWVER=0.74
#
# Initial SIMPLE IP Masquerade test for 2.4.x kernels
# using IPTABLES.
#
# Once IP Masquerading has been tested, with this simple
# ruleset, it is highly recommended to use a stronger
# IPTABLES ruleset either given later in this HOWTO or
# from another reputable resource.
#
#
#
# Log:
# 0.74 - the ruleset now uses modprobe vs. insmod
# 0.73 - REJECT is not a legal policy yet; back to DROP
# 0.72 - Changed the default block behavior to REJECT not DROP
# 0.71 - Added clarification that PPPoE users need to use
# "ppp0" instead of "eth0" for their external interface
# 0.70 - Added commented option for IRC nat module
# - Added additional use of environment variables
# - Added additional formatting
# 0.63 - Added support for the IRC IPTABLES module
# 0.62 - Fixed a typo on the MASQ enable line that used eth0
# instead of $EXTIF
# 0.61 - Changed the firewall to use variables for the internal
# and external interfaces.
# 0.60 - 0.50 had a mistake where the ruleset had a rule to DROP
# all forwarded packets but it didn't have a rule to ACCEPT
# any packets to be forwarded either
# - Load the ip_nat_ftp and ip_conntrack_ftp modules by default
# 0.50 - Initial draft
#

echo -e "\n\nLoading simple rc.firewall version $FWVER..\n"


# The location of the iptables and kernel module programs
#
# If your Linux distribution came with a copy of iptables,
# most likely all the programs will be located in /sbin. If
# you manually compiled iptables, the default location will
# be in /usr/local/sbin
#
# ** Please use the "whereis iptables" command to figure out
# ** where your copy is and change the path below to reflect
# ** your setup
#
IPTABLES=/sbin/iptables
#IPTABLES=/usr/local/sbin/iptables
DEPMOD=/sbin/depmod
MODPROBE=/sbin/modprobe


#Setting the EXTERNAL and INTERNAL interfaces for the network
#
# Each IP Masquerade network needs to have at least one
# external and one internal network. The external network
# is where the natting will occur and the internal network
# should preferably be addressed with a RFC1918 private address
# scheme.
#
# For this example, "eth0" is external and "eth1" is internal"
#
#
# NOTE: If this doesnt EXACTLY fit your configuration, you must
# change the EXTIF or INTIF variables above. For example:
#
# If you are a PPPoE or analog modem user:
#
# EXTIF="ppp0"
#
#
EXTIF="eth1"
INTIF="eth0"
echo " External Interface: $EXTIF"
echo " Internal Interface: $INTIF"


#======================================================================
#== No editing beyond this line is required for initial MASQ testing ==


echo -en " loading modules: "

# Need to verify that all modules have all required dependencies
#
echo " - Verifying that all kernel modules are ok"
$DEPMOD -a

# With the new IPTABLES code, the core MASQ functionality is now either
# modular or compiled into the kernel. This HOWTO shows ALL IPTABLES
# options as MODULES. If your kernel is compiled correctly, there is
# NO need to load the kernel modules manually.
#
# NOTE: The following items are listed ONLY for informational reasons.
# There is no reason to manual load these modules unless your
# kernel is either mis-configured or you intentionally disabled
# the kernel module autoloader.
#

# Upon the commands of starting up IP Masq on the server, the
# following kernel modules will be automatically loaded:
#
# NOTE: Only load the IP MASQ modules you need. All current IP MASQ
# modules are shown below but are commented out from loading.
# ===============================================================

echo "----------------------------------------------------------------------"

#Load the main body of the IPTABLES module - "iptable"
# - Loaded automatically when the "iptables" command is invoked
#
# - Loaded manually to clean up kernel auto-loading timing issues
#
echo -en "ip_tables, "
$MODPROBE ip_tables


#Load the IPTABLES filtering module - "iptable_filter"
# - Loaded automatically when filter policies are activated


#Load the stateful connection tracking framework - "ip_conntrack"
#
# The conntrack module in itself does nothing without other specific
# conntrack modules being loaded afterwards such as the "ip_conntrack_ftp"
# module
#
# - This module is loaded automatically when MASQ functionality is
# enabled
#
# - Loaded manually to clean up kernel auto-loading timing issues
#
echo -en "ip_conntrack, "
$MODPROBE ip_conntrack


#Load the FTP tracking mechanism for full FTP tracking
#
# Enabled by default -- insert a "#" on the next line to deactivate
#
echo -en "ip_conntrack_ftp, "
$MODPROBE ip_conntrack_ftp


#Load the IRC tracking mechanism for full IRC tracking
#
# Enabled by default -- insert a "#" on the next line to deactivate
#
echo -en "ip_conntrack_irc, "
$MODPROBE ip_conntrack_irc


#Load the general IPTABLES NAT code - "iptable_nat"
# - Loaded automatically when MASQ functionality is turned on
#
# - Loaded manually to clean up kernel auto-loading timing issues
#
echo -en "iptable_nat, "
$MODPROBE iptable_nat


#Loads the FTP NAT functionality into the core IPTABLES code
# Required to support non-PASV FTP.
#
# Enabled by default -- insert a "#" on the next line to deactivate
#
echo -en "ip_nat_ftp, "
$MODPROBE ip_nat_ftp


#Loads the IRC NAT functionality into the core IPTABLES code
# Require to support NAT of IRC DCC requests
#
# Disabled by default -- remove the "#" on the next line to activate
#
#echo -e "ip_nat_irc"
#$MODPROBE ip_nat_irc

echo "----------------------------------------------------------------------"


echo -e " Done loading modules.\n"


#CRITICAL: Enable IP forwarding since it is disabled by default since
#
# Redhat Users: you may try changing the options in
# /etc/sysconfig/network from:
#
# FORWARD_IPV4=false
# to
# FORWARD_IPV4=true
#
echo " Enabling forwarding.."
echo "1" > /proc/sys/net/ipv4/ip_forward


# Dynamic IP users:
#
# If you get your IP address dynamically from SLIP, PPP, or DHCP,
# enable this following option. This enables dynamic-address hacking
# which makes the life with Diald and similar programs much easier.
#
echo " Enabling DynamicAddr.."
echo "1" > /proc/sys/net/ipv4/ip_dynaddr


# Enable simple IP forwarding and Masquerading
#
# NOTE: In IPTABLES speak, IP Masquerading is a form of SourceNAT or SNAT.
#
# NOTE #2: The following is an example for an internal LAN address in the
# 192.168.0.x network with a 255.255.255.0 or a "24" bit subnet mask
# connecting to the Internet on external interface "eth0". This
# example will MASQ internal traffic out to the Internet but not
# allow non-initiated traffic into your internal network.
#
#
# ** Please change the above network numbers, subnet mask, and your
# *** Internet connection interface name to match your setup
#


#Clearing any previous configuration
#
# Unless specified, the defaults for INPUT and OUTPUT is ACCEPT
# The default for FORWARD is DROP (REJECT is not a valid policy)
#
echo " Clearing any existing rules and setting default policy.."
$IPTABLES -P INPUT ACCEPT
$IPTABLES -F INPUT
$IPTABLES -P OUTPUT ACCEPT
$IPTABLES -F OUTPUT
$IPTABLES -P FORWARD DROP
$IPTABLES -F FORWARD
$IPTABLES -t nat -F

echo " FWD: Allow all connections OUT and only existing and related ones IN"
$IPTABLES -A FORWARD -i $EXTIF -o $INTIF -m state --state ESTABLISHED,RELATED -j ACCEPT
$IPTABLES -A FORWARD -i $INTIF -o $EXTIF -j ACCEPT
$IPTABLES -A FORWARD -j LOG

echo " Enabling SNAT (MASQUERADE) functionality on $EXTIF"
$IPTABLES -t nat -A POSTROUTING -o $EXTIF -j MASQUERADE



echo -e "\nDone.\n"

[IronBits]

That worked!!! :D
:notworthy

On to the next step, to auto start the clients :)

Edit
Done! Tomorrow I fire up the other 5 PXE nodes :D (which are now running on win98)
Dyyryath contacted me via his Zaurus and will be back tomorrow to tidy up behind me :|party|:

Thank you SO much PCZ and rsbriggs and Dyyryath, for all your HELP getting this working!!!
:notworthy :notworthy :notworthy :elephant:

[rsbriggs]

I've got two of the M7VIQs running PXE boot now, one ASRock, and one old 800 Mhz slot-A using one of the PXE boot cards you recommended. Might try to post a link to some pictures tomorrow...

Had just one minor accident today - wife decided to vacuum my computer room. She found the one open plug in the room, plugged in the vacuum, hit the on-switch, and tripped the breaker.... UPS kept my two servers up, but other 8 comps were out for a while. She understands that she needs to use an outlet in some other part of the house now...

Computer room stays at about 85F even with central air running 24x7. Of course the REST of the house is at about 65F....

[rsbriggs]

Done! Tomorrow I fire up the other 5 PXE nodes (which are now running on win98) Does that end up meaning that you'll then have 5 extra/spare hard drives? That'll be a nice reduction in power and heat... :thumbs:

I've observed that the AMDs crunch DF a little faster running Win-XP than they do running Linux, and Intels crunch a little faster running Linux. Somewhat curious... :confused:

[IronBits]

Have you figured out where the MAC address is marked on the VIQ mobo?
I looked around and didn't see one. Not a big deal...
Didn't figure it would be in the manual
/me shrugs and mumbles something about the 'hard way' -
boot up in PXE mode tail /var/log/messages, edit dhcpd.conf and restart dhcpd ;)

Vacuum around computers? :scared:

Yes, I will be able to unplug 5 40GB HDDs tomorrow :D
Each HDD uses about .1 amps, but puts out alot of heat for sure...

Are you using the icc client? It's faster than the gcc version.

BTW, this is all running under Mandrake 9.1 :thumbs:

I documented alot of what I did, but there is the part that Dyyryath did I have to wait for, then I'll replilcate the effort on another Server to make sure I have it down pat, then, I'll write a script that just puts it all out there where it belongs, with default cfg files, with instructions where to edit and for what reasons, then I'll tar it all up in one file and make it available as one downloadable file, untar it, run the script and away it goes ;) (well, I like to dream a little)
I'll prolly need some help with the scripts ...

Still need to work out the shutdown problem!
Can't shutdown from the console of the node, nor with telnet yet :(

Then there is the shell script that monitors all the nodes progress.txt file and puts it in HTML format so you can just hit a webpage on the PXE Server to see if they are all running :)

[PCZ]

IB
Glad you have finally worked it out.
You create an SMB share on the server and use DCmonitor.

[IronBits]

SMB ? Is that secure?
A fancy perl script and a cron job to run it every 15 minutes would do fine ;)

[Paratima]

In Linux, ifconfig gets your MAC addresses displayed.

It's shown as HWaddr...

[IronBits]

Nice thinking, but, these are headless pxe boot nodes ;)

[PCZ]

I monitor all the clients from a machine on my lan which is privately addressed not on the internet.

Most folks will have machines sitting on private address's not on publicly routable
IP's.

SMB shares work fine and DCmonitor is an excellent tool to monitor multiple clients.
You can set up a web server to use the HTML output from DCmonitor if you want to.

You have chosen to give your server a public address so security is an issue for you.
It sounds like you are planning to make your stats publicly accessable.

[Paratima]

Originally posted by IronBits
Nice thinking, but, these are headless pxe boot nodes ;) Telnet works, yes? Or ssh, yes? These headless things are not brainless, yes?

[IronBits]

They can't boot up until the MAC address is known, so there is no OS ;)
but wait, they do have win98 running right now...
Great idea! :hifi:
/me runs off to VNC each node :D

[magnav0x]

Glad to hear you got things going IB! I haven't checked up on this thread in a long time. Guess I'll reconsider those Biostar motherboards after all :cheers:

[IronBits]

These BIOSTAR M7VIQ (http://www.newegg.com/app/viewproduct.asp?DEPA=1&submit=Go&description=BIOSTAR%2CM7VIQ) are only $55 and with onboard vid/pxe boot nic, they are a nice edition to the pharm, and, they are GREAT performers. (266 FSB only - great for XP2400)
PCZ has had good success with a different brand to, but don't know the cost or performance...
rsbriggs has both, maybe he could comment...

BTW, check your signature, df counter no longer works since mr_mann left ;)
Maybe we could get Dyyryath to do the df counter thing instead!! :D

[rsbriggs]

I end up booting each one twice when I first bring them up. Once, and I tail /var/log/messages to find their MAC address, edit dhcpd.conf and lts.conf, then reboot and watch var/log/messages again to make certain that they connected.

Do you have a simple way to set up a Samba share, PCZ? I'm not concerned about security on the farm sub-net - there are two sets of NAT translations and 3 separate firewalls between them and the internet.

But, I'll tell you what, setting up ZoneAlert on my Windows DSL router box was certainly eye-opening. I think someone must have been running a porn server or something on my system, given the number of hits (now starting to dwindle down) I was getting on port 80, and that is supposed to be AFTER the USWest NAT firewall.

Also, after contacting the Mandrake Linux site, I'm getting hits on port 6881 from ezoffice.mandrakesoft.com about once per second. Wonder what THAT is all about?

The M7VIQ boards are quite easy to deal with. My only concern about the boards is that there are some electrolytic caps on the board that are running at close to 200 degrees F. - somewhat over their spec'ed temperature rating. The caps run much hotter than the processor or either of the bridges....

[magnav0x]

Download samba and install it. Of course this will all depend on what distro you are using. I use gentoo, so all I do is 'emerge samba'. You guys may be able to use RPM's or may be forced to download the source, build and install it. Replace your smb.conf with this and of course edit it to share the directory of your choosing. For example, I made the user 'pcguest' with a password of null and put all the stuff I wanted to share out (like my df client for monitoring from my windows box). This will work right away, assuming you have the 'pcguest' user on your box with no password. After you have replaced the smb.conf with this just issue the 'smbd -D' command to start the samba daemon. Good luck, hope it helps.

[global]
workgroup = workgroup
server string = none
interfaces = 192.168.1.1/64 (variable)
security = SHARE
encrypt passwords = YES
password level = 2
username level = 2
log file = /var/samba
max log size = 50 (default)
lock dir = /lock
guest account = pcguest
[tmp]
comment = (your coments here)
path = /usr/tmp
read only = No
directory mask = 0775
guest ok = yes
browseable = no

[homes]
comment = Home Directories
browseable = no
read only = no
create mode = 0750

[printers]
comment = All Printers
browseable = no
printable = yes
public = no
writable = no
create mode = 0700

[Gentoo Samba]
comment = pcguest
path = /home/pcguest
public = yes
writable = yes
printable = no
write list = @staff

[rsbriggs]

Gave it a try, but ended up with the "usual" result - can't see the share from any Windows boxen. Never have been able to get Windows boxes to see Linux boxes successfully on this network. I'd LOVE to be able to do it, since Textpad on Windows is my preferred editor in any and all situations.

[Dyyryath]

Samba's pretty simple, I use it extensively both at work and at home.

Couple of questions (which may have already been answered farther up, but I'm feeling lazy ;)):

(1) Are you using a domain controller or are your machines peer to peer?

(2) What's the domain name/workgroup name you're using? If you're using a domain, what's the PDC's name?

(3) Are they all on the same subnet?

(4) What distro are you using?

(5) What version of Windows are your other boxes using?

Given these, I can set you up a config that will work fine.

[IronBits]

1) peer to peer
2) dbestern
3) yes - LAN side
4) Madrake 9.1
5) XP or w2k

Is it done yet? (over here on ether silly) :rotfl:

[rsbriggs]

1) peer to peer
2) MSHOME
3) no, but that isn't the problem. The ONE important machine is dual homed
4) Red Hat 9.x
5) XP-Pro

[magnav0x]

rsbriggs, I can't "see" my linux boxes either. But if you goto start->run and type \\192.168.1.102 or whatever the IP for the linux box is, it SHOULD connect.

[rsbriggs]

\\192.168.0.2
The network path was not found

ping 192.168.0.2
no problem

[the-mk]

Originally posted by rsbriggs
\\192.168.0.2
The network path was not found

ping 192.168.0.2
no problem

is there a firewall blocking port #139?

[rsbriggs]

Nope - it's on the same subnet. I read the samba man page for a while, and tweaked. Still can't get NMBD to run for some reason, so can't "see" the computer, but a \\IP direct connect and some tweaks results in this DCMonitor output:



CN Computer name Struct Gen Next gen Buff gens Best energy Status
4 Kathey 0 0 0 0 0.000 Unknown 09:45 31/08
1 Dell4600-1 0 0 0 0 0.000 Unknown 09:45 31/08
12 Barton-1 21 54 29 1 6.190 Running
6 OldAMD 41 62 9 1 5.651 Running
2 Dim4500 6 106 44 0 5.864 Running
5 Dell600SC 11 126 39 0 6.282 Running
3 Dell4600-2 31 138 19 1 4.511 Running
13 Linux DF-1 1 184 49 0 10000000.000 Running
14 Linux DF-2 16 153 34 1 4.875 Running
15 Ws002 6 33 44 1 6.418 Running
16 ws003 21 244 29 1 4.046 Running
17 ws004 26 208 24 1 4.065 Running
18 ws005 6 130 44 0 4.096 Running


(dunno where computer # 7, 8, 9 or 10 are :dunno: )

[PCZ]

rsbriggs

When you installed red hat you basically did a custom install and selected everything
Based on this assumption you will have the same tools as me installed.

Creating an SMB share on the red hat server takes a few seconds only and you dont have to touch smb.conf

You already have windows PC's in either a workgroup or domain.

Your windows PC's log on with a usermame and password. the password might be blank but it is still a password



from the start menu in KDE/Gnome select System Settings , Server Settings , Samba Server.

A little applet will load called Samba Server Configuration.

1:From the Preferences menu select Server settings
Type the name of your workgroup or the netbios name of your domain in the workgroup box

click on the security tab and make sure Authentication mode is set to User
Click OK.

3: From the Preferences menu select Samba Users.
click on the add user button.
The Create New Samba User dialog box will pop up.
click on the drop down to the right of Unix Username and select root from the menu.
fill in the windows username and password boxes.
click ok. and ok again

4:
Click on the green add button
The Create Samba Share dialog box will pop up

Browse to the directory you want to share.
Type a description if you want.

Change the setting of Basic Permissions to Read /Write

Click on the Access button
it should have Only allow access to specific users selected
pick a tick in the box called root
click ok

select File Quit

the config will be saved and Samba restared automatically.

Note: This config is only for use in a secure enviroment.

[rsbriggs]

Unfortunately, I made the mistake of installing Ximian Evolution, which has me trapped inside a Gnome login, and seems to restrict me from MANY selections I had available before installing it. Let me re-try a log-on as root instead of as a normal user.

EDIT:

Nope. Anyone know how I can get back to my nice KDE desktop? I can't find where a display manager is being set in Red Hat...

DCMonitor needs a column for "structures changed" on a computer by computer basis, rather than just totalling it....

[PCZ]

I think the html that it outputs could be fed into a stats system.
Over to someone else to work that one out.

[IronBits]

www.dbestern.com/dcmonitor.htm :D
Dyyryath is busy working some of his magic :notworthy

[PCZ]

so how do you get the best energy and status in the web output

[AMD_is_logical]

I've used Biostar M7VIT pro motherboards in a diskless cluster (with mii.o and via-rhine.o drivers). They don't have onboard video, but there is a "Halt on no errors" option that lets them run without a video card (once they're set up). They can run fsb=333MHz. There's no power-on after power-fail option (that I could find), so I've been putting a 100uF cap where the power-on switch would normally go. That causes it to turn on when the power goes on.

[IronBits]

Originally posted by PCZ
so how do you get the best energy and status in the web output I dunno, I'm using your suggestion of a Samba share and DCmonitor. ;)

[IronBits]

Originally posted by AMD_is_logical
There's no power-on after power-fail option (that I could find), so I've been putting a 100uF cap where the power-on switch would normally go. That causes it to turn on when the power goes on. Nice! Where are the docs and pics of course? :D

[PCZ]

Sorted
I was using version 0.7
version 0.72 has the extra info

[IronBits]

Funny how a little thing like a .02 can make ;)
Glad you got it running.
I moved em all over to pxe nodes, lost one in the shuffle...
Might be a little bead of sweat that got on it, so will let it dry out and try again in a few hours. :rolleyes:
/me shrugs
Glad that's over with :D

[PCZ]

I bet you know a lot more about linux now ?

I found it quite a steep learning curve but well worth it.

[IronBits]

Well... I know a lot more about pxe for sure. ;)
Thanks to you, rsbriggs and Dyyryath :D :thumbs:
:cheers:

[AMD_is_logical]

Originally posted by IronBits
Nice! Where are the docs and pics of course? :D You want a pic of the cap?
http://jnw0.home.mindspring.com/cap.jpg

[IronBits]

Nice pic! :thumbs:
That's all it takes to get a motherboard to boot up if it doesn't support power on when electricity is restored?!
I like it :)

[Paratima]

Schweet! :)

[AMD_is_logical]

Originally posted by IronBits
That's all it takes to get a motherboard to boot up if it doesn't support power on when electricity is restored?!
I like it :) I've only tried it with Biostar M7VIT pro motherboards, but it works great with them. :thumbs: