[8.04] Mythbuntu-diskless: Testers wanted!

[laga]

Hello,

Mythbuntu 8.04 will make it very easy to add diskless frontends to your existing setup.

There are several reasons why you'd want diskless frontends:
* No hard disk: your frontend is cheaper and quieter
* Setup time: you don't have to install Mythbuntu on all your frontends. Instead, you just create one image on the server which is used by all diskless frontends.
* Quickly turn existing (windows) computers into MythTV frontends without having to install Mythbuntu or load the live disk.

The second version of the mythbuntu-diskless packages was uploaded last night and they're finally stable enough to have them tested by a wider audience.

If you're interested in testing it, I have put together some documentation (https://help.ubuntu.com/community/MythTV/Install/Hardy/Diskless). It's necessary that you're at least running mythbuntu-diskless version 0.2-0ubuntu1.

Keep in mind that it's a bit rough, especially regarding the documentation :). If you encounter any problems or if you have trouble understanding the documentation, please post here so we can get it fixed!

Regards,

Michael

update #1: There seems to be some confusion whether mythbuntu-diskless uses fat clients or thin clients.
With thin clients, all applications are run on the server and the thin client just displays the graphical parts. For MythTV, this doesn't make much sense because transferring uncompressed video over your network will use a lot of bandwidth.
That's why we're using a fat client approach: everything is running locally on the client. You should be able to access your optical drives, hard disks and capture cards just like with a regular Mythbuntu install. (Keep in mind that mythtv-backend isn't automatically installed, though.)

update #2: the documentation now resides in the wiki. Please help to improve it! https://help.ubuntu.com/community/MythTV/Install/Hardy/Diskless

[rickyble]

Thanks any help is appreciated. I started down the usb disk and actually looked like it was saving everything to that location including the img. The usb was a freebie a vendor had given up and it was on 64mg and so slow. Anyway got impatient and removed and did look at the content. I had no tv/video card in that laptop. I have since ordered on from ebay to use. I think this weekend I will get a bigger usb stick and try it again when the card comes in. In the

[laga]

rickyble,

you're probably in the wrong thread here. This is exclusively for testing/bug reports for mythbuntu-diskless. If you're having issues with the live disk, I'm sure someone can answer in your original thread (eg if you can come up with a reproduceable test case).

[venom986]

holy crap, ignore me, ltsp not lstp! yeesh.

[venom986]

Ok, this was pretty quick and easy up to this point. I followed your docs, quickly edited the /etc/ltsp/dhcpd.conf to match my network topology and the fact that i am using amd64, not i386.

I'm now at the point where the client successfully negotiates and boots over PXE, and I get the Mythbuntu logo and progress bar. However, the progress bar does its little dance back and forth for a bit and then freezes. This seems to happen every time (have tried 3 times). I am now trying to see if there is any log output to direct me to what the problem is.

[venom986]

Feb 29 10:24:08 cerebro mountd[22435]: refused mount request from 192.168.1.3 for /var/cache/mythbuntu-diskless/overlay/ (/): not exported


Looks like the culprit. Did I miss instructions on exportin the overlay directory?

[laga]

Yes, you probably missed the dpkg-reconfigure step. If you didn't, can you show me your /etc/exports (and the output of exportfs as wel)?

[venom986]

I ran dpkg-reconfigure, but received no prompts.
/etc/exports

/home/storm/var 192.168.1.3(rw,async,no_root_squash,no_subtree_che ck)
/media/sda 192.168.1.1/24(rw,async,no_root_squash,no_subtree_check)
/media/md0 192.168.1.1/24(rw,async,no_root_squash,no_subtree_check)
/media/md1 192.168.1.1/24(rw,async,no_root_squash,no_subtree_check)
/media/md2 192.168.1.1/24(rw,async,no_root_squash,no_subtree_check)


exportfs

/home/storm/var
192.168.1.3
/media/sda 192.168.1.1/24
/media/md0 192.168.1.1/24
/media/md1 192.168.1.1/24
/media/md2 192.168.1.1/24

[laga]

Odd.

What version of mythbuntu-diskless-server do you have? 0.2* is the minimum required version.

If the version is correct, could you please try sudo dpkg-reconfigure --priority=low mythbuntu-diskless-server?

[Calash]

This is probably an obvious question,

I am running 0.21 on Feisty 7.10. Can I follow the instructions to install this or do I need to upgrade to 8.04?

[laga]

This is probably an obvious question,

I am running 0.21 on Feisty 7.10. Can I follow the instructions to install this or do I need to upgrade to 8.04?

Good question. It's basically possible if you backport a few packages from hardy: ltsp, debootstrap and the squashfs utils (I don't remember the exact name). The client itself can only use Hardy however so you'll have to add --dist hardy to your ltsp-build-client line.

Take care if you're already using the squashfs utils somewhere. It's possible that you won't be able to open images generated with the backported squashfs-utils with your current kernel. Just a heads-up; it doesn't matter for mythbuntu-diskless.

I'm running (and developing) mythbuntu-diskless on my Gutsy box so it's certainly possible :)

[venom986]

hmm. it appears that since I had previously installed it, my attempt to grab the latest this morning with apt-get install mythbuntu-diskless-standalone failed to get all the latest stuff. Just closed down mythfrontend on the primary backend server and had notices about new packages, inlcuding the new diskless.

installing it now, and will retry the dpkg-reconfigure. I won't be able to try getting the diskless frontend working again until monday though.

I presume I will have to run that update script to update the compressed filesystem that the client will be booting into?

[venom986]

Okay, I have the correct export now. should get back to you monday on success of booting the client.

what needs to be done to keep the client environment updated as package updates become availalbe? do i have to chroot in periodically to run the updates?

[Calash]

I have been thinking it over and I think I will do the full distro upgrade to Hardy, then try it out. I already got 0.21 running fine, so it will mostly just be background upgrades.

Thanks for the info :)

[laga]

what needs to be done to keep the client environment updated as package updates become availalbe? do i have to chroot in periodically to run the updates?

Yes. Mythbuntu-control-centre will have a button to start synaptic inside the chroot, though.

It's *necessary* that you upgrade your chroot after installing the new packages.

After upgrading, run update-initramfs -k all -c inside the chroot to make the initramfs pick up the new scripts. Afterwards, run sudo ltsp-update-kernels outside the chroot.
Or just start over :)

[venom986]

are you talking about running mythbuntu control center from the client machine?

Also, I was going to suggest you add those instructions for synching updates the chroot to your documentation; but when I went to check if you already mention it I found that the site you linked to for documentation is down.

[venom986]

Well, here's a brief update. I had to reboot my frontend and didn't cancel out of the PXE boot quick enough; so, it successfully booted over the network ;)

However,

1) It booted to a login prompt rather than mythfrontend
2) When I logged in on it I had a desktop and it didn't launch mythfrontend
3) The network manager applet thought it didn't have an IP address. considering this was a successful NETboot, that seems erroneous ;)

[laga]

Well, here's a brief update. I had to reboot my frontend and didn't cancel out of the PXE boot quick enough; so, it successfully booted over the network ;)


Great :)


1) It booted to a login prompt rather than mythfrontend
2) When I logged in on it I had a desktop and it didn't launch mythfrontend


Yes, that's intentional. Just launch mythbuntu-control-centre and enable auto-login there.
Maybe automatic login should be enabled, though. If there's an easy way to do that while building the chroot, I might just add it. Until just now, I hadn't realized that it's a problem :)


3) The network manager applet thought it didn't have an IP address. considering this was a successful NETboot, that seems erroneous ;)

Yes, that's a problem. We're currently telling HAL to ignore all network devices. Before we did that, HAL/network-manager/something else would reconfigure the network device and that's really nasty if you do NETboot :) Maybe we should just disable Network-Manager?

Can you please test if the power button works on your diskless frontend?

[laga]

are you talking about running mythbuntu control center from the client machine?


You can run mythbuntu-control-centre on the client, but it's not recommended to install any packages there. I was talking about a change to MCC where you can update the chroot on the host system.


Also, I was going to suggest you add those instructions for synching updates the chroot to your documentation; but when I went to check if you already mention it I found that the site you linked to for documentation is down.

I'll put the whole documentation in the Wiki instead.

[venom986]

You can run mythbuntu-control-centre on the client, but it's not recommended to install any packages there. I was talking about a change to MCC where you can update the chroot on the host system.

Has that change to the MCC gone live yet? I didn't see an option when I checked. Then again, I think my Primary Backend is still waiting for a reboot after the last update and I've had recordings going that i didn't want to interrupt so haven't rebooted yet.

[laga]

Has that change to the MCC gone live yet? I didn't see an option when I checked.

No, mythbuntu-control-centre doesn't know anything about diskless yet. I'll update the link to the documentation now, maybe you want to add your experiences whereever you see fit: https://help.ubuntu.com/community/MythTV/Install/Hardy/Diskless

[venom986]

I'll definately add my experiences once I get a chance to sum them up.

I guess until mythbuntu-control-centre gets updated, its probably easiest to nuke the client image and create a new one?

how does this handle things like remotes or display drivers that are installed on the client itself rather than updated in the chroot?

[laga]

I guess until mythbuntu-control-centre gets updated, its probably easiest to nuke the client image and create a new one?


You can just do your changes in the chroot and run ltsp-update-image when you're done, like i described earlier. Redoing the image is easier, but iz'll take more time :) You'll also not have to worry about updating the initramfs then.


how does this handle things like remotes or display drivers that are installed on the client itself rather than updated in the chroot?

I'd like to make a distinction between changing settings and installing new software.

* If you change a setting on the client, it'l be written to the overlay directory (/var/cache/mythbuntu-diskless/overlay/$key on the server). If you reboot the client, it'll still be available.

* The same applies when you install new software. However, it's not recommended to do that. When you use apt-get to install new software, the list of installed packages etc on your system is updated. Under the hood, this is done by aufs by copying the file that is to be modifed into the overlay directory where any changes are applied.
The client now has its own copy of the package list. But what's going to happen if you install new packages into the chroot and ltsp-update-image afterwards? The files provided by those packages will most likely be available to the client (unless they were created before on the client), but apt-get on the client doesn't know that there are new packages installed.

So, the big problem here is that the more changes you make, the more your client will become out of sync with your original squashfs. It's not a big problem during normal operation, but it should be kept in mind. Imagine someone upgrades their systems to the next Ubuntu release and apt-get on the client still thinks it's on the last version. But you can also run into problems when an applications changes their config file format and the client still has a deprecated version.

You can easily remove offending files in /var/cache/mythbuntu-diskless/overlay/$key, though.

The only case I can think of where you should install software on the client is software like nvidia-glx, eg when you have clients with different generations of Nvidia cards and each client needs their own version of nvidia-glx.

[venom986]

I'd like to make a distinction between changing settings and installing new software.

* If you change a setting on the client, it'l be written to the overlay directory (/var/cache/mythbuntu-diskless/overlay/$key on the server). If you reboot the client, it'll still be available.

* The same applies when you install new software. However, it's not recommended to do that. When you use apt-get to install new software, the list of installed packages etc on your system is updated. Under the hood, this is done by aufs by copying the file that is to be modifed into the overlay directory where any changes are applied.
The client now has its own copy of the package list. But what's going to happen if you install new packages into the chroot and ltsp-update-image afterwards? The files provided by those packages will most likely be available to the client (unless they were created before on the client), but apt-get on the client doesn't know that there are new packages installed.

So, the big problem here is that the more changes you make, the more your client will become out of sync with your original squashfs. It's not a big problem during normal operation, but it should be kept in mind. Imagine someone upgrades their systems to the next Ubuntu release and apt-get on the client still thinks it's on the last version. But you can also run into problems when an applications changes their config file format and the client still has a deprecated version.

You can easily remove offending files in /var/cache/mythbuntu-diskless/overlay/$key, though.

The only case I can think of where you should install software on the client is software like nvidia-glx, eg when you have clients with different generations of Nvidia cards and each client needs their own version of nvidia-glx.

Plus, I suppose, if you do the updates in the chroot then you only have to do them once and they become available to all your diskless clients.

I look forward to getting back to setting this up tomorrow.

[venom986]

so, i can get things to boot on the client. however, i've yet to convince my calls to aptitude upgrade in the chroot to actually apply kernel updates. this means when i loaded into my client I had to install them from there.

next issue, in many cases, when using mythbuntu control center to enable various options, I get an error regarding installArchives()

[laga]

Why doesn't upgrading in the chroot work? Error message?

Can you also provide the error messages from mythbuntu-control-centre?

Btw, there are some pending updates which will disable networkmanager and the update-notifier completely, thanks for reminding me about that :)

[venom986]

aptitude upgrade simply reported the kernel updates as held back for some reason

the errors from mythbuntu-conrol-centre center around it trying to run configure on various mythtv plugins. it was popping up an error with instalArchives() in the box.

I seem to be having issues with getting nvidia restricted drivers to work. i end p in Low Graphics Mode when I reboot the client. I think this may be a mismatch between kernel version and restricted drivers package version though.

I ended up going into chroot and into aptitude's ncurses gui to force the kernel updates. have just rebooted again.

the other issue i'm having (and haven't confirmed yet if its been fixed by upgrading the kernel versions) is that I can't nfs mount my media shares off the server. I get an unkown filesystem type message from mount.

[venom986]

the mounting problem seems to be that /sbin/mount.* don't exist on the client.

chrooting in and running aptitude install portmap nfs-common to try to fix that aspect of things.

oh, and the errors surrounding the mythplugins exist when doing that too.

dpkg: dependency problems prevent configuration of mythflix:
mythflix depends on mythtv-frontend (>= 0.20-0.0); however:
Package mythtv-frontend is not configured yet.
dpkg: error processing mythflix (--configure):
dependency problems - leaving unconfigured
dpkg: dependency problems prevent configuration of mythvideo:
mythvideo depends on mythtv-frontend (>= 0.20-0.0); however:
Package mythtv-frontend is not configured yet.
dpkg: error processing mythvideo (--configure):
dependency problems - leaving unconfigured
dpkg: dependency problems prevent configuration of mythbrowser:
mythbrowser depends on mythtv-frontend (>= 0.20-0.0); however:
Package mythtv-frontend is not configured yet.
dpkg: error processing mythbrowser (--configure):
dependency problems - leaving unconfigured


that's just an excerpt

[venom986]

okay, after having installed portmap and nfs-common in the chroot and updated my image, i booted back into the client.

my shares mount but i get:
sudo mount 192.168.1.5:/media/sda /media/sda
Can't set permissions on mtab: Operation not permitted

[venom986]

Okay, so it appears that /etc/fstab resets on every reboot? how do set up my nfs mounts in a persistent way?

[laga]

Wow, I'm sorry to hear that it gives you so much trouble.

With regard to fstab, there's no easy way I can do it from the package right now. I'll either have to patch LTSP or just refrain from installing LTSP stuff in the chroot and move that logic to the mythbuntu-diskless-client package.

For now, I'd suggest you edit /etc/init.d/ltsp-client-setup and comment out the configure_fstab call in line 242 (and rebuild your image or just do it on the client).

[venom986]

on another note, i have the feeling a LOT of my problems stem from the fact that the users in the client system do NOT have the same UIDs as the primary backend

[laga]

on another note, i have the feeling a LOT of my problems stem from the fact that the users in the client system do NOT have the same UIDs as the primary backend

That''d only give you problems with NFS. Or did you mean something else?

[venom986]

Wow, I'm sorry to hear that it gives you so much trouble.

With regard to fstab, there's no easy way I can do it from the package right now. I'll either have to patch LTSP or just refrain from installing LTSP stuff in the chroot and move that logic to the mythbuntu-diskless-client package.

For now, I'd suggest you edit /etc/init.d/ltsp-client-setup and comment out the configure_fstab call in line 242 (and rebuild your image or just do it on the client).

No worries, its still alpha ;)

I tried editing the fstab file in the chroot but that didn't persist either. guess I'll comment out that line in the ltsp-client-setup (will it then take what i have in the chroot?). Hmm, for that matter, I could edit the configure_fstab function.

In fact, what do you think about having that function source another file that would contain custom mounts? that would allow the user to specific any additional mounts that their system needs without mucking in the script. You could even allow for two files, one that applies to ALL clients, and one that is client specific?

[venom986]

That''d only give you problems with NFS. Or did you mean something else?

I'm getting permissions errors trying to view some of my media from the client. basically when it crosses through to the nfs mounted filesystem the UID no longer matches when checking permissions.

[laga]

In fact, what do you think about having that function source another file that would contain custom mounts?

I've just talked to the LTSP guys and they've made configure_fstab() configurable (yay!). It'll take a few days for that change to go into Ubuntu, though.

If the user wants to have mounts on all clients, he'll have to add them to the chroot.. of course, that won't work if the clients already have their own fstab :/

[laga]

I'm getting permissions errors trying to view some of my media from the client. basically when it crosses through to the nfs mounted filesystem the UID no longer matches when checking permissions.

Yes, that's a NFS problem and there's nothing I can do about (ie not caused by mythbuntu-diskless). Back then, when I had that problem, I just created a "mediafiles" group with the same GID on both the server and the client. That's not perfect either but it helps. There's probably a HOWTO somewhere which knows better than I do :) NFS version 4 probably can solve your problem.. nfs-user-server also has basic UID mapping abilities, but it can't transfer files bigger than 2G which makes it utterly useless.

[laga]

I think I've found out why you had trouble installing the nvidia driver. The linux-restricted-modules package was not installed :/ I'll fix it in the LTSP plugin.

[Calash]

The new Mythbuntu-control-centre has this as an option now. Tested it out last night and it built the image without a problem. The interface could use a bit of work, I still have not figured out what the bottom part of the screen is for.

I cannot test until I get a DHCP server setup, but I have a laptop ready for this once I do.

Quick question. I found this article and was wondering if this type of setup would work for this. I have a Netgear router and really do not want all the house computers dependant on my Mythbuntu box. If I am reading it correctly I can set it up this way and any system requesting the image will get it's IP from the Mythbox

http://www.etherboot.org/wiki/twodhcpservers

[laga]

Hi!

Before you try to boot your clients, you need to run
sudo dpkg-reconfigure mythbuntu-diskless-server
on the server to set up the NFS export for the overlay directory. This step will be automated in MCC soon.


The new Mythbuntu-control-centre has this as an option now. Tested it out last night and it built the image without a problem. The interface could use a bit of work,


If you have any suggestions, please let me know. It's pretty much impossible for a "newbie" to use mythbuntu-diskless without having read the documentation first, though. And I have yet to write that documentation. :) (Which doesn't mean you shouldn't try, just ask and I'll answer and compile the documentation from this thread later ;)).



I still have not figured out what the bottom part of the screen is for.

I cannot test until I get a DHCP server setup, but I have a laptop ready for this once I do.


The bottom part of the screen is exactly for that problem - it's one of several possible solutions actually. If you've got a spare USB pen drive and your client can boot from USB, you can create a bootable pen drive there. You need to select your diskless environment (usually 'i386'), your network interface so the client knows where it can find the server and of course your pen drive.

It'll copy the kernel and the initramfs to the first partition of the pen drive and set up a bootloader. It expects that your pen drive is formatted with fat32. You probably have to set that partition to 'bootable' in cfdisk. If you don't know how to use cfdisk or fdisk, just ask.. or at least don't overwrite your hard disk :)


I have a Netgear router and really do not want all the house computers dependant on my Mythbuntu box. If I am reading it correctly I can set it up this way and any system requesting the image will get it's IP from the Mythbox

http://www.etherboot.org/wiki/twodhcpservers

Yes, that should work. I think I used a similar setup a few years ago when I was first messing with diskless clients (on Gentoo back then ;)). I believe the key to getting this working is the following sentene (from your link):



At least the recent etherboot versions ignore DHCP offers without “filename” (and your netgear router probably does not send out filename if not especially configured to do so).

It probably doesn't work with other PXE implementations which might accept DHCP offers without the necessary information for booting, so make sure you use etherboot.


There's a third option:
you can set up your DHCP server to use alternate DHCP ports. You'll also have to use a special version of etherboot which you can easily get using their rom-o-matic website. You'll also have to edit /etc/initramfs-tools/conf.d/ltsp inside the chroot and set the DHCPPORT variable accordingly. Then update your initramfs (inside the chroot) using update-initramfs -k all -c and then run sudo ltsp-update-kernels outside the chroot.
To make your dhcpd run on a different port, you have to edit the init script and add that option there.


Phew, another piece of documentation written ;) Maybe you'd like to try some approaches and give me some feedback.. even the link you posted would be fine so we know what works and what doesn't.

[Calash]

I will run through some of the options tonight with my test laptop. It is i386, and has a good chunk of memory so it should be fine.

I would love to find a way to make this work wireless. I know it may be a bit slow, but I have read some posts on a PVR forum where a few people got HD content to stream wireless.

One step at a time.....wired first :)

[laga]

Wireless is not a big problem, although you won't be able to use PXE for that. You can't really boot over network when the network connection hasn't been established yet :)

You'll have to resort to the pen drive method and you'll have to include a initramfs script which connects the interface to the wireless network. I think I have found such a script while googling around, so you should be able to do the same :)

[Calash]

On the subject of the interface, I took another look and it is more user error than a problem with it. My eyes did not catch the bold black letters separating the USB Flash thumbdrive from the Image Maintenance area, so they merged together in my head.

Last night I tried a couple of things with various degrees of success.

USB Drive boot did not work and got stuck in a loop of mount errors. I can get the actual errors later.

I installed the DHCP part of the diskless server package and had an interesting result. With my current setup (Netgear router has DHCP on, as well as my Myth box) I was able to boot into the image via PXE, but it got stuck after the Mythbuntu load screen. It looks like it had set the ROOTDRIVE to my router, but I did not diagnose it further. Tonight I will play with the Myth DHCP config settings and see if I can get it sorted out.

[laga]

On the subject of the interface, I took another look and it is more user error than a problem with it. My eyes did not catch the bold black letters separating the USB Flash thumbdrive from the Image Maintenance area, so they merged together in my head.

Last night I tried a couple of things with various degrees of success.

USB Drive boot did not work and got stuck in a loop of mount errors. I can get the actual errors later.



Yes, I'd appreciate some error messages. If it was some kind of loop, it probably happened because it couldn't mount the squashfs image.. does /opt/ltsp/images/i386.img exist? Is it listed in /etc/inetd.conf? Is inetd running?

[BustedChicken]

I decided to post my diskless adventures in here instead of creating a new thread out there.
I followed the instructions and now the client boots up. The problem now is, I get to the Mythbuntu startup logo, The kernel then goes haywire and panics.

[ 50.119559] Kernel panic - not syncing: Attempted to kill init


There is nothing else on the screen, just that. I did have a somewhat complicated dhcp.conf, though I removed it to install the simple version that was recommended and changed it to reflect the amd64 arch.

Both machines are identical, in fact, once this is running, all of the machines will be identical; 4 clients total and 2 servers. They all run amd64 processors. The servers have 2 gigs of ram and the clients have 4 gigs.

[laga]

I followed the instructions and now the client boots up. The problem now is, I get to the Mythbuntu startup logo, The kernel then goes haywire and panics.

[ 50.119559] Kernel panic - not syncing: Attempted to kill init



Can you go to /var/lib/tftpboot/ltsp/amd64/pxelinux.cfg/ and remove "quiet splash" in the "default" file? Then boot your client and post the output. Thanks!

Nice boxes, btw :)

[BustedChicken]

...
Negotiation: ..size = 417960KB
bs=1024, sz=417960
Done.
mount call failed: 13
mount: Mounting /cow/001a925a0612 on root/cow failed: Invalid argument
Begin: Running /scripts/nfs-bottom ...
Done.
Done.
Begin: Running /scripts/init-bottom ...
Done.
run-init: overmounting root: Stale NFS filehandle
[ 48.561721] Kernel panic - not syncing: Attempted to kill init!


I don't remember mounting cows...huh?

Where is the log for this, or is it not saved on the server?

[laga]

mount call failed: 13 tells me that the NFS server doesn't allow your client to mount the overlay directory. Running sudo dpkg-reconfigure mythbuntu-diskless-server will fix this (you'll have to answer some questions).

'cow' is short for copy-on-write, btw ;) I'm going to add that dpkg-reconfigure magic to mythbuntu-control-centre now.

[BustedChicken]

mount call failed: 13 tells me that the NFS server doesn't allow your client to mount the overlay directory. Running sudo dpkg-reconfigure mythbuntu-diskless-server will fix this (you'll have to answer some questions).

'cow' is short for copy-on-write, btw ;) I'm going to add that dpkg-reconfigure magic to mythbuntu-control-centre now.

It didn't work. Same error. I am wondering if it wasn't something I did earlier. If I uninstall diskless-server using apt, would it give me a clean slate to retry, or do I need to clean some stuff up?

[laga]

It didn't work. Same error. I am wondering if it wasn't something I did earlier. If I uninstall diskless-server using apt, would it give me a clean slate to retry, or do I need to clean some stuff up?

I'd prefer if you could take a look at /etc/exports and check if you have a line similar to:
/var/cache/mythbuntu-diskless/overlay/ *(rw,no_root_squash,async,no_subtree_check)
in your /etc/exports. That 'error 13' usually indicated that there was a permission problem with the nfs server for me so you should be really close to have a working setup. What version of mythbuntu-diskless-server do you have, btw?

[BustedChicken]

Edit: Just realized half my post is missing.
Needless to say, this is in /etc/exports:


/var/cache/mythbuntu-diskless/overlay/ *192.168.1.0/24(rw,no_root_squash,async,no_subtree_check)


The deb that I downloaded was mythbuntu-diskless-server_0.4-0ubuntu2_all.deb

Hope this helps

[laga]

/var/cache/mythbuntu-diskless/overlay/ *192.168.1.0/24(rw,no_root_squash,async,no_subtree_check)


The deb that I downloaded was mythbuntu-diskless-server_0.4-0ubuntu2_all.deb

Hope this helps


Remove the asterisk: * in front of the net mask (either using your favourite editor or using sudo dpkg-reconfigure mythbuntu-diskless-server) and reload the nfs-kernel-server. It should start working afterwards.

[BustedChicken]

Success, thanks for your help. I will provide some feed back after I 'drive' it for a day or 2.

I have several machines to hook up, I assume this is pretty much going to be plug and play.

[Calash]

After some work last night I am able to do the following.

PXE boot to Mytbbuntu Desktop as long as the Mythbuntu box is the only active DNCH server on the network.

USB flash boot to Mythbuntu with two DHCP servers (Netgear and Mythbuntu), however the Myhbuntu DHCP has to be active or the USB boot does not work.

At this point I can open Mythfrontend, but it will not connect to my Backend, no matter how I boot. The control centre shows the MySQL info as valid, and from the laptop I can connect to it no problem via command line, but the terminal gives the following error.


Connecting to backend server: 127.0.0.1:6543

Connection timed out.
You probably should modify the Master Server
settings in the setup program and set the
proper IP address



127.0.0.1 is not what I entered into any of the myth setup screens, and it is not what is in the MySql.txt file, so I am a bit stumped as to why it is trying localhost instead of remote (192.168.0.4)


Almost there :)

[laga]

USB flash boot to Mythbuntu with two DHCP servers (Netgear and Mythbuntu), however the Myhbuntu DHCP has to be active or the
USB boot does not work.


That's odd. I'll have to try that. Does it give you an error message?


127.0.0.1 is not what I entered into any of the myth setup screens, and it is not what is in the MySql.txt file, so I am a bit stumped as to why it is trying localhost instead of remote (192.168.0.4)


Can you open mythtv-setup on the master backend and make sure that both IP addresses in the setup screen are not set to 127.0.0.1?

[Calash]

Ah-HA....knew it was something simple. Changed the settings on the master backend and everything fired up great.

I will try to get you some error messages from the USB next. Thanks for all your work on this :)



Edit: Just tested again and it made a liar out of me :). Worked fine with Mythbuntu DHCP disabled. Sorry for the false alarm :)

[Calash]

Well, I have been testing over the past couple of days and here is what I have found. Some of this is due to a non-standard config that I will explain below.


My Mythbuntu box has 3 drives in it. The first is the OS drive, then a second 80gb drive is mounted as /storage in the root. This drive holds my Video and Music directories. Finally a third drive is under that, mounted as /storage/Record. This is where my recordings are held.

The diskless front end boots fine with both PXE boot and USB boot. I can watch live TV fine, but by default I cannot watch and videos, music, or watch recordings. As far as I can tell this is due to my setup more than a bug in the system. For videos all I had to do was mount /storage/Video and everything worked.

For music I have to mount /Storage/Music, then go change the setting to point the music directory to that location and rescan the collection. This got everything working fine.

Recordings are still being a pain. The console shows it is trying to open files in that directory, but failing. It was late last night so I did not get to complete my diagnosis of this, but I will post once I get more info.

I also noticed two other oddities. The first is that I cannot open MCC by pressing the button in the "parent" MCC. It errors out saying something about not being authenticated. This may be related to a host file entry I have on my computer that lets me map the Mythbox as 'linuxserv'. This does not exist on the Mythbox, so it may be causing problems. Not a big deal at this point as I can just use the terminal to run updates.

The other oddity was during the updates it had me update settings for Mythweb. Should mythweb be installed on a remote front-end? In the MCC of the frontend is shows Mythweb as being checked, but greyed out so I can not edit it.

Other than that it words great. My test system is a laptop and it has no problem with connecting and playing TV.

[laga]

Calash,

do you have other (working) remote frontends or is this your first try? Could you post some logs from the client? See /var/log/mythtv/

[jerryscuba]

I'm not sure if this has been reported yet...
I tried installing Mytbuntu 8.04 alpha 4 Desktop version on a Epia board with 256MB RAM. After working forever on the partitioning the installer ran out of memory at 15% progress and just ended without any messages. I guess installing needs more memory than running mythtv...

[laga]

I'm not sure if this has been reported yet...
I tried installing Mytbuntu 8.04 alpha 4 Desktop version on a Epia board with 256MB RAM. After working forever on the partitioning the installer ran out of memory at 15% progress and just ended without any messages. I guess installing needs more memory than running mythtv...

I think you got the wrong thread here..

[suprslug]

I got my thin client to boot. But don't understand how to setup my music and video. Everytime I go to setup anything the frontend crashes. Mythstream works, but shouldn't it recognize my backend setup? If it's connected to the backend, why can't I watch tv or videos or listen to music?

[laga]

I got my thin client to boot. But don't understand how to setup my music and video. Everytime I go to setup anything the frontend crashes. Mythstream works, but shouldn't it recognize my backend setup? If it's connected to the backend, why can't I watch tv or videos or listen to music?

See the link in my signature to find out how to provide log files.

If it crashes, you should be asked to file a bug report. Please do that. If you're not asked: does /etc/rc2.d/S20apport exist on the client? Is apport-gtk installed on the client?

[suprslug]

When it crashes I get no errors at all. It'll even take down firefox with it. No bug report. I do have /etc/rc2.d/S20apport in /opt/ltsp/i386/etc/ folder and apport-gtk installed.

[laga]

Does it crash the whole computer? Can you a bit more specific? You can also find logs in /var/cache/mythbuntu-diskless/overlay/*/

[suprslug]

Now the fronetend won't even start. I'm going to trash /opt/ltsp and start fresh. It's nice to get the sound going over to a thin client. So, I might as well rebuild the image and start fresh. I'll post what happens.

[laga]

You should also remove the overlay directory of your client in /var/cache/mythbuntu-diskless/overlay/. You'll loose all of your settings for your client that way.

[bgrolleman]

Ok, so I finally had time to rebuild my old mythtv setup with mythbuntu, and everything is working fine on the server and with a live CD on another desktop. :-D

Current Setup,
Mythbuntu backend server and diskless server (.10)
WRT54GL router with Tomato firmware (.3)
Diskless Client, via-epia board

So, to start the PXE boot, my router is running dnsmasq uses these settings,

dhcp-vendorclass=pxe,PXEClient
dhcp-vendorclass=etherboot,Etherboot
dhcp-option=17,/opt/ltsp/i386
dhcp-boot=net:pxe,/ltsp/i386/pxelinux.0,,192.168.54.10
dhcp-boot=net:etherboot,/ltsp/i386/nbi.img,,192.168.54.10


After removing the quiet splash tags, I could see that it all worked fine until the system tries to mount the NBD, simply because it looks at my router (.3) and not the mythbuntu server (.10)

I tried the dual DHCP setup and that works, but I'm still wondering, where to set the NBD server?

[laga]

Hi,



dhcp-vendorclass=pxe,PXEClient
dhcp-vendorclass=etherboot,Etherboot
dhcp-option=17,/opt/ltsp/i386
dhcp-boot=net:pxe,/ltsp/i386/pxelinux.0,,192.168.54.10
dhcp-boot=net:etherboot,/ltsp/i386/nbi.img,,192.168.54.10



I don't know where your problem lies exactly, but here's what works for me on my wrt54g:


dhcp-boot=ltsp/i386/pxelinux.0,prometheus,192.168.0.137


The host name 'prometheus' resolves to 192.168.0.137.

If I had to guess, I'd blame"dhcp-option=17,/opt/ltsp/i386" for your trouble..


I tried the dual DHCP setup and that works, but I'm still wondering, where to set the NBD server?

Edit /var/lib/tftpboot/ltsp/i386/pxelinux.cfg and add "nbdserver=ADDRESS:PORT". Note that this config file will be overwritten when you run ltsp-update-kernels, so you should use a different configuration file. Check the pxelinux documentation, it'll tell you which config files are valid.

You can find the port in /etc/inetd.conf. You still might want to try that line I posted from my dnsmasq.conf, it works for me.

[bgrolleman]

Excelent, trimming down my DNSMASQ configuration worked.
I also tried the nbdserver option, but this didn't have any effect.

So on my router the extra dnsmasq configuration only has this,
dhcp-boot=ltsp/i386/pxelinux.0,tools-ws,192.168.54.10

and that works great.

Now I still get a black screen in the end of booting, but since it does the same using the live CD I will post that in another thread.

[laga]

Now I still get a black screen in the end of booting, but since it does the same using the live CD I will post that in another thread.

That's a shame. Here's some diskless-specific information: you can edit the xorg.conf on the server, it's in /var/cache/mythbuntu-diskless/overlay/<overlay-dir>/etc/X11/

[chukdotcom]

I'm having problems with my /etc/fstab. I modified it on the server. The diskless clients get the correct file. However, my NFS mounts do not load at startup. If I do a "sudo mount -a" after it boots, they load fine. Am I missing something?

[laga]

Are your shares mounted if you run sudo /etc/init.d/mountall.sh start on the client?

[chukdotcom]

No they are not. What is the proper way to add my own mount? I would like all the diskless clients to share that same mount.

[laga]

The proper way is adding it to /etc/fstab, it's just possible that it doesn't work because the appropriate init script isn't used.

If it doesn't work after you run '/etc/init.d/mountall.sh start' on the client, then I'm stumped. Maybe you can try it again?

[chukdotcom]

Well, I'm a little confused. If I make changes to the /etc/fstab on the client, they are lost when I re-boot. If I make them in the client environment on the server, they are ignored. I had to edit /etc/init.d/ltsp-client-setup in the client environment on the server. In this section:

configure_fstab() {
if [ -z "$CONFIGURE_FSTAB" ] || boolean_is_true "$CONFIGURE_FSTAB" ; then
echo "/dev/root / unionfs defaults 0 0" > /etc/fstab
echo "tmpfs /tmp tmpfs defaults,nosuid,nodev 0 0" >> /etc/fstab
mount /tmp
fi


I changed it to:

configure_fstab() {
if [ -z "$CONFIGURE_FSTAB" ] || boolean_is_true "$CONFIGURE_FSTAB" ; then
echo "/dev/root / unionfs defaults 0 0" > /etc/fstab
echo "tmpfs /tmp tmpfs defaults,nosuid,nodev 0 0" >> /etc/fstab
echo "192.168.25.5:/var/lib/mythtv/videos /mnt/videos nfs rw,user,rsize=8192,wsize=8192,exec,hard 0 0" >> /etc/fstab
mount /tmp
fi


That /etc/fstab makes it down to each of the clients, but my "custom" mount doesn't mount. I also tried throwing in the line "mount /mnt/videos" after the "mount /tmp" line.

[chukdotcom]

Also, where can I see if any of this mounting stuff is logged?

[laga]

I can see why your fstab is overwritten.. the lts.conf file is not set up properly.

Take a look at /opt/ltsp/i386/etc/lts.conf. CONFIGURE_FSTAB needs to be changed to CONFIGURE_FSTAB=false. After you've done that change, you can just copy the file to /var/lib/tftpboot/ltsp/i386/lts.conf as indicated by the instructions in that file.

Does it work now? I'm still not sure if it's actually mounted, but at least your fstab shouldn't be overwritten anymore.

[ddempsey3]

Is anybody having issues updating the client using the Control Centre? I can successfully build a client but I am not able to update.

When I select the image and select Control Centre I successfully get the Control Centre for the diskless client but when I commit the changes, the CPU goes to 100% and the Control Centre just hangs.

My arch is amd64 for both server and client. Any ideas?

[ddempsey3]

I realized that I can update config, like auto login, but I am not able to install new packages.

The download seems to be successful but when the install begins python takes 100% of the CPU and never finishes.

[laga]

The download seems to be successful but when the install begins python takes 100% of the CPU and never finishes.

You're right. I'll take a look.

[TheSideways]

Hi,
This seems only half-off topic, but I really couldn't find anything about it anywhere else. What hardware works good as a diskless front end? I like the idea, keep a small, quiet box next to the TV. Has anyone built a box for this purpose? Do the general mythtv guidelines apply, just some extra ram because of no local swap space? Has anyone built a passive-only cooled box for this? That's my number one complaint about mythtv boxes -- the noise. I'll try to check back here, but I'd apreciate it if you could cc answers to thesideways at google's mail server.
Thanx,
Tim

[pharnet]

This is a popular subject on silentpcreview.com (http://www.silentpcreview.com). I would start with their reference section (http://www.silentpcreview.com/section5.html) and then move on to the forums.

I personally am uncomfortable with a totally passive pc, I think some airflow is wise. Some of the low power consumption cpus like AMD's Sempron LE(45w) or dual core Athlon BE(45w) or the Intel Celeron 420(35w) run much cooler so you can turn down the speed of the fan(s). In my latest quiet machine I did the following:

Used the Sempron LE-1100 and got rid of the cpu fan by using the Scythe Ninja Plus RevB CPU Cooler passively. Make sure this cooler will fit in your case and on your motherboard, it is big.

Got rid of the power supply fan by using the picoPSU 90 (http://www.mini-box.com/picoPSU-90-power-kit?sc=8&category=13). It is an extremely small, efficient, fanless power supply.

For the only fan, I used one of SilentPCReviews Reference case fans in my case the 92mm Nexus Real Silent Case Fan and turned down its speed to a comfortable level. If my case used a 120mm fan I could have used the fan that came with the Ninja.

I prefer to stick to mATX motherboards and cases since they are mass market priced and easy to find locally. The Ninja will be the item that constrains your choice.

I was using an old laptop hard drive but it died recently and it is fortuitous that this project exists. Diskless is the way I will be going. Thanks Laga and everyone else for doing it. If I did replace the hard drive I would use 2.5" laptop SATA model since they use 1 to 5w and are quiet and cool.

I had a PXE boot environment already so it was easy for me and I have encountered no issues so far with Mythbuntu Diskless.

[laga]

I had a PXE boot environment already so it was easy for me and I have encountered no issues so far with Mythbuntu Diskless.

Great! Thanks for your nice feedback :)

[laga]

Hi,

there was another LTSP upload last night which had some changes for the mythbuntu-diskless plugin. If someone feels like testing, make sure you've got LTSP 5.0.40~bzr20080212-0ubuntu4 installed and recreate your chroot.

This should also fix any problems with mounting NFS shares etc as all init scripts are now enabled.

edit: It'd also be good if you wiped the overlay directory when testing (or move it out of the way). Of course, if you do that, you'll lose any customizations like your lircrc. You can find the overlay directory in /var/cache/mythbuntu-diskless/overlay/.

[Dilligaf]

I just found this thread and I'm trying to get this working. I'm running Hardy with all updates and MythTv installed from Mythbuntu control center. I went into control center and had it create an image, after creating an image I click on Control Center in client maintenance and get this "Sorry, I couldn't find mythbuntu-control-centre in the chroot!" Shouldn't I be able to open control center? I'm running an old computer with Endian Firewall as a router/dhcp server and am capable of editing the dhcpd.conf file as needed, can someone who has this working off an external dhcp server post their dhcpd.conf file for comparison/guidance ? I've read through this thread and absorbed most of it but the directions are installing from the command line, is that still necessary or should the GUI handle it now? Any help or guidance will be greatly appreciated.

Mike

Edit: After clicking on the terminal button I can browse chroot and after searching I can't find any MythTV binaries, what am I missing?

[laga]

I went into control center and had it create an image, after creating an image I click on Control Center in client maintenance and get this "Sorry, I couldn't find mythbuntu-control-centre in the chroot!" Shouldn't I be able to open control center?

[..]

Edit: After clicking on the terminal button I can browse chroot and after searching I can't find any MythTV binaries, what am I missing?

It's possible that something went wrong while building the chroot. There were some updates last night (LTSP and mythbuntu-control-centre), can you try deleting and re-creating the chroot (if your bandwidth permits)?

If you have any custom repos in your /etc/apt/sources.list, you need to check the "accept unsigned packages" checkbox.

Let me know how it goes.

In case you want to do some experimentation or if your net connection is slow, it might be a good idea to set up apt-cacher.

[Calash]

Do we still download the LTSP via WGET at this point?

I should have time to work with this tonight.

[laga]

Hi Calash,

no. Everything should work out of the box now. Make sure you have LTSP 5.0.40~bzr20080212-0ubuntu5, there was another upload last night.

[laga]

I've started to put together some documentation. Maybe someone wants to take a look and make some suggestions or even contribute something.

You can get the pdf at http://laga.ath.cx/complete-manual.pdf (11.22MiB) or the source at https://code.launchpad.net/~mythbuntu/mythbuntu/documentation (look for en/diskless.tex).

Yes, I know that one of the screenshots is updated, we'll update them all in one go :)

[Dilligaf]

I tried deleting and rebuilding the image this morning and still had the same problem, no myth binaries. I noticed an update came out today, I updated and tried to delete the image and Mythbuntu Control Center crashed. I manually deleted the image and attempted to rebuild, Mythbuntu CC crashed again when hitting reply. Thanks for any help.

Mike

[Calash]

I have been having odd problems with MCC as well. It let me delete the image, but it ignored my attempts to make a new one. No error in the log, just ignored it.

I buiilt the image in the terminal and it worked well, but seems to have some other problems. I had no sound and no mouse on my test laptop. LiveTV played fine, and I was able to play recordings with no problem.

I will dig around a bit and see if there is anything interesting in the logs.

[craiga]

same thing here laga

choose amd64, checked allow unsigned packages presed build and....nothing. the button just stayed depressed, this is on a fresh install.

building the client via the terminal which seems to work fine as the above poster said

[laga]

Hi,

I kind of expected that this was going to be an usability issue :)

You need to click "apply" after pressing the button.

Make sure you got the latest control centre (0.27-0ubuntu1), there was another upload last night which also fixed the problem where the image couldn't be deleted.

[Calash]

Apply does nothing. I got one error about dhcp3 in the terminal, but most of the time you click apply and it acts like you selected nothing.

I am going to verify my version of MCC tonight...It seems to be more of a problem with that.

[laga]

The dhcp3 error is fixed for me in the last upload.

I've just fixed the problem where the image wouldn't build correctly in MCC, but I don't know if we can get that uploaded today.. I might just post a fresh .deb here.

[laga]

Hi,

I've uploaded a new mythbuntu-control-centre deb to my webspace:

http:/laga.ath.cx/mythbuntu-control-centre_0.28-0ubuntu1~laga1_all.deb

Changelog:


* private build for testers in the mythbuntu forums
* add MCC_DEBUG environment variable. set to "true" to get additional
debugging output. Right now, only applies for ltsp-build-client
* fix the progress bar for -diskless

[finlay648]

I tried setting diskless up by installing a new server from 8.04 beta and then selecting the Build Image and Apply. I set up my dhcp server and copied the pxelinux.0 file to /var/lib/tftpboot/ltsp/i386 (I was expecting the install process to do this but it didn't. My client has builtin PXE so that's what I'm using. The client boots and retrieves the pxelinux.0 file and boots it but hangs trying to find the pxelinux.cfg configuration file. Where is that file supposed to come from? I assume there should be a default file. Was it supposed to be built during the server image build process? I'm stuck.

Thanks

John

[laga]

Try again with the .deb I posted above.

[finlay648]

I installed the MCC from your link but still doesn't install anything in /var/lib/tftpboot.
Using the ltsp-build-client script does install enough to boot and run the client. I did have some difficulty upgrading the server. I got a message that indicated that only a partial upgrade was possible so I let that continue. Since then it seems the server is fully up-to-date.

John

[Dilligaf]

What I did was copy everything in the boot directory to the root.

Mike

[finlay648]

I assumed that MCC would do all the copying/setup for diskless clients but it doesn't appear that it does yet. I also have a problem with stability running 8.04 - it seems to hang regularly and the display gets hashed. Not sure if it's a hardware or software problem. I think I'll have to abandon this effort until there is a more stable release or I can try it on different hardware.

[laga]

I assumed that MCC would do all the copying/setup for diskless clients but it doesn't appear that it does yet.

Well, it's supposed to do that :)

Maybe you can do me a favor and run the following command:

MCC_DEBUG="true" sudo -E /usr/share/mythbuntu-control-centre/bin/mythbuntu-control-centre


in a terminal. That'll produce lots of debugging output which you can then copy into a pastebin or a text file and post here. Make sure that the scroll buffer in your terminal is large enough.. or just redirect to a file by adding "2>&1 > mylogfile.txt" to the above command.

Then we can figure out what's going wrong :)

[finlay648]

The system being used for the tests has become unstable and hard to get test results from. I did a complete reinstall and tried setting up the diskless environment. This time it seemed to work though the system kept crashing and I had to start over with the image builds. But the /var/lib/tftpboot/ltsp directory was populated and I was able to boot a diskless client. I can only suppose that because of the system instability the previous install was defective and the diskless setup attempt was twarted by that. Thanks for your help and I'll try again after I replace the motherboard.

[encom]

This project is great! However I was having problems right out of the gate with my Asus P5KL-VM (Onboard NIC Attansic L1)

So here is how I got mine fixed.

To get a Attansic L1 Gigabit Ethernet nic card to work:


On the server:


>sudo mount -o bind /proc /opt/ltsp/i386/proc/
>sudo chroot /opt/ltsp/i386

>sudo vi /etc/initramfs-tools/modules

add "atl1"

>sudo update-initramfs -u

>sudo vi /etc/udev/rules.d/70-persistent-net.rules

Add the following two lines between the dashes:
-----------

#add "# PCI device 0x1969:0x1048 (atl1)"
SUBSYSTEM=="net", DRIVERS=="?*", ATTRS{address}=="00:1e:8c:7d:59:e8", NAME="eth0"

-------------

Replacing the appropriate MAC address with your own. To find your MAC address look at your DHCP log.

>exit
>sudo umount /opt/ltsp/i386/proc/
>sudo ltsp-update-image

That worked for me. Of course you can replace atl1 with your own NIC driver that isn't booting at startup.

Hope this helps someone.

[sym_zo]

Hello,

I'm trying to boot the client, and it fails.
The PXE image gets loaded, And I see the mythbuntu splash screen, but then it stops on a black screen with a blinking cursor.
If I switch to console 1 (ctrl-alt-f1), a message is repeated numurous times:
Error : connect : connection refused
(edit)
It happens just after the following line :
filename:/ltsp/i386/nbi.img
(/edit)

It happens before any nfs mount, because after some time the messages stop, and only then can I see this message on the server :
Apr 21 11:50:28 tramp mountd[15088]: authenticated mount request from 192.168.1.140:804 for /var/cache/mythbuntu-diskless/overlay (/var/cache/mythbuntu-diskless/overlay)

Then on the client, a CLI starts (busy box ?) :
(initramfs)

can someone help ?
thanks a lot, and thanks to the contributors for their hard work !!

[sym_zo]

Oops, I solved my problem.

The server needed a restart for nbd-server to work.

It works now ! youhou ;-)

[finlay648]

Well, it's supposed to do that :)

Maybe you can do me a favor and run the following command:

MCC_DEBUG="true" sudo -E /usr/share/mythbuntu-control-centre/bin/mythbuntu-control-centre


in a terminal. That'll produce lots of debugging output which you can then copy into a pastebin or a text file and post here. Make sure that the scroll buffer in your terminal is large enough.. or just redirect to a file by adding "2>&1 > mylogfile.txt" to the above command.

Then we can figure out what's going wrong :)

New MB installed and new 8.04 AMD install. Original problem has reoccurred but this time I have the log file attached as a bz2.

Hope this helps.

[Paul Lowman]

Hi

I am in the throws of installing the diskless client. My first attempt using the MCC failed when several files could not be found by apt. I then succeeded by following the MythTV/Install/Hardy/Diskless document. Unfortunately, when I try to boot the client it gets as far as the TFTP phase and stalls - I guess it cannot connect to the TFTP server. I have checked the dhcp configuration and also inetd.conf and all appears OK. I have successfully used PXE on the client previously so I am sure it is not the problem. Any ideas ?

[laga]

I then succeeded by following the MythTV/Install/Hardy/Diskless document.


Did you download the plugin? Because that's not necessary any more :)

What files could not be found? When? Please be more specific.


Unfortunately, when I try to boot the client it gets as far as the TFTP phase and stalls - I guess it cannot connect to the TFTP server. I have checked the dhcp configuration and also inetd.conf and all appears OK. I have successfully used PXE on the client previously so I am sure it is not the problem. Any ideas ?

Is the tftp server listed in /etc/inetd.conf? Is there anything interesting in /var/log/daemon.log?

[laga]

New MB installed and new 8.04 AMD install. Original problem has reoccurred but this time I have the log file attached as a bz2.

Hope this helps.

Hi,

thanks for the log files. I can't reproduce your problem here. Can you show me your /etc/apt/sources.list file? Also, did you update your Mythbuntu after installing?

[finlay648]

Hi,

thanks for the log files. I can't reproduce your problem here. Can you show me your /etc/apt/sources.list file? Also, did you update your Mythbuntu after installing?


Here's my /etc/apt/sources.list contents:

#deb cdrom:[Mythbuntu 8.04 rc amd64]/ hardy main restricted universe
deb http://archive.ubuntu.com/ubuntu hardy main restricted universe multiverse
deb http://archive.ubuntu.com/ubuntu hardy-updates main restricted universe multiverse
deb http://archive.ubuntu.com/ubuntu hardy-security main restricted universe multiverse
deb http://archive.ubuntu.com/ubuntu hardy-backports main restricted universe multiverse
deb-src http://archive.ubuntu.com/ubuntu hardy-updates main restricted universe multiverse
deb-src http://archive.ubuntu.com/ubuntu hardy universe multiverse main restricted
deb-src http://archive.ubuntu.com/ubuntu hardy-backports main restricted universe multiverse
deb-src http://archive.ubuntu.com/ubuntu hardy-security main restricted universe multiverse

I had forgotten to update my system and did so but had the same result.

Thanks

John

[Paul Lowman]

Hi

I have given up on the original diskless install as I relaised I was using the beta release. I have now installed the release candidate but when I try to install a diskless role I get the message that two files could not be retrieved. This is the same as I was experiencing previously and made me try the manual install method outlined in the Hardy/Diskless document.

Anyhow the two files in question are:

ltsp-server-5.0.40~bzr20080212-ubuntu.all.deb
mythbuntu0diskless-server-0.8-0ubuntu.all.deb

Any clues ?

[finlay648]

I think I figured out what was going on. When building a client image using mcc a couple of ltsp-build-client processes are started and the main mcc window is hidden and a small progress dialog is displayed. When the progress dialog is closed and the main mcc window is displayed I assumed that the client build process had been completed. But it seems that the two child processes continue to run. It takes about 3m additional before the /var/lib/tftpboot/ltsp/<arch> directories are created. But it appears that the child processes hang during the creation of the /opt/ltsp/images/<arch>.img file and don't complete until mcc is quit. After mcc is quit the processes continued to run for another 10m finally finishing the creation of the .img file.

Originally I was confused by this behavior and with some of the system problems that were occurring the system often crashed before completing the client build even though I thought it had completed when the main mcc window reappeared. I think that better feedback in the GUI would help a lot.

I believe that there is something preventing the completion of the client build process until mcc is shut down.

[laga]

But it seems that the two child processes continue to run.

Only one ltsp-build-client process is supposed to be running - you probably started it twice.

Anyways, this should be fixed in the deb I've posted above. Did you use that? If not, maybe you can install it and try again?

Thanks for the testing!

[laga]

Anyhow the two files in question are:

ltsp-server-5.0.40~bzr20080212-ubuntu.all.deb
mythbuntu0diskless-server-0.8-0ubuntu.all.deb

Any clues ?

You need to update your package lists. There is a button at the bottom of the screen in the Control Centre for that - I can't look up the name right now. Or just use "sudo aptitude update".

Thanks for the feedback!

[finlay648]

Only one ltsp-build-client process is supposed to be running - you probably started it twice.

Anyways, this should be fixed in the deb I've posted above. Did you use that? If not, maybe you can install it and try again?

Thanks for the testing!

You're right there is only one ltsp-build-client process the other was the sh -c that invoked it. I thought the above .deb was a special debug version of the mcc that is in 8.94rc so I didn't download it. But I see it has fixed the problem of the lbc process hanging.

Why is the client hostname some long random number instead of the name assigned using in dhcpd,conf?

Thanks

john

[laga]

You're right there is only one ltsp-build-client process the other was the sh -c that invoked it. I thought the above .deb was a special debug version of the mcc that is in 8.94rc so I didn't download it. But I see it has fixed the problem of the lbc process hanging.


I should have made that clearer :) The above deb is what will be uploaded into the archives soon (if it hasn't been uploaded already). Is it working for you now? Great!


Why is the client hostname some long random number instead of the name assigned using in dhcpd,conf?


It's the MAC address of your client :) That's the only unique identifier I could think of, it's also used to assign the overlay directory to the client.

You should be able to get the normal hostname back by setting the $HOSTNAMEOVERRIDE variable in the initramfs to something different from "true". Either add "hostnameoverride=blah" to your boot arguments or edit the initramfs config file. Changing the boot arguments is probably the easiest method. I've written some documentation last night:

http://bazaar.launchpad.net/~mythbuntu/mythbuntu/documentation/annotate/laga%40moar-20080423100953-4xu9cyo8uaxsgz8p?file_id=diskless.tex-20080410183931-h3pnflfppvayqrju-1 around line 209

I realize it's awkward to have to read the latex source directly, but we currently have some problems with pdf prevention. I'm sure you can read it if you squint your eyes a little bit :)

[finlay648]

So I noticed a couple of bugs in mcc:

1. can't apply the change after unchecking the automatic login checkbox.
2. spelling error "advanced mangement" should be "advanced management"

Is that something you can deal with or should I file a bug elsewhere?

Otherwise it seems to work i.e. booted up and contacts the server, etc. Thanks for doing this - I was planning on trying to figure out how to do diskless clients and then I saw that you already had done it.

[Paul Lowman]

Hi

Updating the packages worked in as far as the files are now found. However although the Diskless role did load some files the full diskless environment is not installed - no tftp, no client image no ltsp - what am I missing ? Do I have to use the manual method still?

Sorry to keep bothering you but I am really looking forward to getting this all going :)

Cheers

[laga]

Hi

Updating the packages worked in as far as the files are now found. However although the Diskless role did load some files the full diskless environment is not installed - no tftp, no client image no ltsp - what am I missing ? Do I have to use the manual method still?

Sorry to keep bothering you but I am really looking forward to getting this all going :)

Cheers

Did you use the .deb I posted above?

[malbrigt]

Works good when following wiki instructions, my laptop ( Intel C2D ) works great as a network booted frontend.

However when booting on my intended frontend, a small-form Via Epia V series mainboard, I receive "illegal instruction" while trying to excecute mythfrontend.

Seems like optimized binaries, anyone know for sure architecture of diskless image really is i386?

[laga]

Yes, the binaries are optimized. You should be able to recompile them. Maybe you can put them on a PPA so everyone can benefit. Or we might just do that.. someone on these forums already has a PPA for epia-specific builds, but I'm not sure if those binaries will work on your CPU or if it's compiled for a newer VIA CPU.

[malbrigt]

Thanks for quick response.

I'll try to find those, epia is a popular platform for frontends so it would be cool if there was a default option for building compatible images.

Anyone who know about a good information resource regarding custom compiling mythtv on mythbuntu and integrating it to the diskless image?

[Paul Lowman]

Hi Laga

I installed the deb you poster earlier and the client image built OK over the weekend but it still seems as though the tftp configuration is missing. I can see no pxelinux.0 file or a pxe config file for the clients under pxelinux.cfg - am I still being dumb or is there something I still need to do?

At the client it gets through the dhcp stage OK but the tftp reports file not found. My understanding is that the tftp gets its file name from the config files after running pxelinux.0

Cheers

[laga]

It'd help if you posted the complete error message. :)

Anyways, it seems something went wrong while creating the chroot :( One of the last steps is the creation of /opt/ltsp/i386/etc/lts.conf - can you show me the content of that file?

Also, do the missing configuration files show up if you run "sudo ltsp-update-kernels" outside the chroot?

[Verzweifler]

Hi there,

I do face an issue right now when it comes to configuring the diskless clients using an instance of MCC in the chroot.
After building the image (i386 that is) I am just not able to start the MCC using the the GUI button as suggested by the guidance. Every time I try to start it, an error pops up telling me that there is no MCC found in the chroot. Did I miss something? I think I followed the instructions from the Installation Manual word-by-word... :confused:

[laga]

What version of mythbuntu-control-centre are you using? Use "apt-cache policy mythbuntu-control-centre" to find out. Thanks!

[Verzweifler]

What version of mythbuntu-control-centre are you using? Use "apt-cache policy mythbuntu-control-centre" to find out. Thanks!

Issuing

apt-cache policy mythbuntu-control-centre yields the following output

mythbuntu-control-centre:
Installed: 0.27-0ubuntu1
Candidate: 0.27-0ubuntu1
Version table:
*** 0.27-0ubuntu1 0
500 http://archive.ubuntu.com hardy/universe Packages
100 /var/lib/dpkg/status
I did an upgrade to 8.04 yesterday, but it also did not work before...

[laga]

Verzweifler,

you'll need to get MCC from the hardy-proposed repository. Sorry for that inconvenience, but that last-minute fix just couldn't make it in.

If you've (successfully) tested that version 0.28 from hardy-proposed works, please comment on https://bugs.launchpad.net/ubuntu/+source/mythbuntu-control-centre/+bug/221921 so it can go into hardy-updates.

Again, sorry for the inconvience, but things don't always go as smooth as I'd like them to go.

BTW: on the Mythbuntu 8.04 live disk, we've already shipped 0.28. Those people don't have to worry about the update.

[Paul Lowman]

Laga

I did not send the exact error message as it simple said TFTP ERROR File not found ...

There is no /opt/ltsp/i386/etc/lts.conf file. Also the sudo ltsp-update-kernels did not produce any files ...

The /var/lib/tftboot/ltsp and i386 directories are both empty ...

I will download the 8.04 live iso and try again ...

[laga]

Just get the package from hardy-proposed, that should be less hassle..

[teqdruid]

I'm trying to get myself running with mythbuntu-diskless. Looks like I'm having trouble on my client connecting to nbd. I didn't have inet running at first, but once I started it up, I started getting this error at boot:

Negotiation: ..size = 330592KB
bs=1024, sz=330592
<4>sysfs: duplicatte filename 'pid' can not be created
WARNING: at /build/buildd/linux-2.6.24/fs/sysfs/dir.c:424 sysfs_add_one()
Pid: 6146, comm: nbd-client Not tainted 2.6.24-16-generic #1

Followed by a kernel stack trace, then an error mounting /rofs since /dev/nbd0 doesn't exist. Eventually I get to a prompt, and when I run "nbd-client 192.168.2.10 2000 /dev/nbd0" I get the same error message.

I typed the error message off the screen, so there might be a typo in it, and I didn't have the patience to type the stack trace.

Any hints?

root@neon:~# apt-cache policy mythbuntu-diskless-server
mythbuntu-diskless-server:
Installed: 0.9-0ubuntu1
Candidate: 0.9-0ubuntu1
Version table:
*** 0.9-0ubuntu1 0
500 http://archive.ubuntu.com hardy/universe Packages
100 /var/lib/dpkg/status

TIA,
John

[teqdruid]

I'm trying to get myself running with mythbuntu-diskless. Looks like I'm having trouble on my client connecting to nbd. I didn't have inet running at first, but once I started it up, I started getting this error at boot:


So, I probably should have noted that I was using a flash disk boot. I finally got PXE boot working with my router, and it booted no problem. Anyway, problem solved.

[sjrice]

I'm trying to generate an etherboot image from Rom-O-Matic (to burn to a CD -- I have an old machine that doesn't support USB booting). I don't want to use DHCP, so I'm using the USE_STATIC_BOOT_INFO option. What I'm stumped on is what path to use the STATIC BOOTFILE option. I've tried using /opt/ltsp/images/i386.img as the path, but it doesn't work. It tries to download by tftp and fails.

By the way, has anyone figured out a way to boot the diskless image from a usb stick with wireless network support? Being able to use Mythtv wirelessly on my work laptop without installing anything would be the cat's meow!

[laga]

I'm trying to generate an etherboot image from Rom-O-Matic (to burn to a CD -- I have an old machine that doesn't support USB booting). I don't want to use DHCP, so I'm using the USE_STATIC_BOOT_INFO option. What I'm stumped on is what path to use the STATIC BOOTFILE option. I've tried using /opt/ltsp/images/i386.img as the path, but it doesn't work. It tries to download by tftp and fails.


Try: ltsp/i386/pxelinux.0

Once you've got it working, do you think you could post a quick HOWTO in this thread so I can add it to the docs? That'd be a great contribution.



By the way, has anyone figured out a way to boot the diskless image from a usb stick with wireless network support? Being able to use Mythtv wirelessly on my work laptop without installing anything would be the cat's meow!

You just need a initramfs script to set up the wireless connection. It shouldn't be too hard. I've found this German web site: http://www.lug-kr.de/wiki/ThinClientLokalBooten

Maybe you can use some of their scripts (and extend them for wpa_supplicant support). I was going to work on wireless support, but it just didn't happen for this release.

[flyguy99]

I upgraded my mythbackend from 7.10 to 8.04 and pulled mcc from hardy-proposed. However, I still get the "sorry, I couldn't find mythbuntu-control-center in the chroot" message. Did I still need to follow the steps outlined for the mcc prior to .28?

[Verzweifler]

Try this thread... ;-)

http://ubuntuforums.org/showthread.php?t=775489 (http://ubuntuforums.org/showthread.php?t=775489)

After some investigation, laga (thanks again, pal!) provided me with an updated .deb that resolved my issue. That .deb will make it to hardy-proposed but until then, you could just use what he attached. It is in one of the last posts...

[flyguy99]

Thanks! you guys are awesome!:)

[mondo1287]

I can't seem to get etherboot to work with any nic on an old Athlon XP board.

I'm running DHCP on a Windows 2003 box and it is configured properly for PXE mythbuntu booting. DHCP is giving out the tftp server address, root path, and file name. I've succesfully tested it with a VMWare client using VMWare's built in PXE support, using an Etherboot floppy image with VMware, and I've also tested it with another machine using its nic's built in PXE support. Everything seems to be working great with those machines.

I've tried several different nics with the problem machine using an etherboot floppy and I can't get any of them to work. They are all unable to obtain a DHCP lease. The nics' link lights, including the motherboard's built in nic, never light up. Etherboot appears to detect the card but then cannot obtain a lease. Also, the switch does not show there is a link either. The nics do work if the machine is booted into an OS, so I think the nics themselves, cables, and the switch can all be ruled out. Etherboot seems to be unable to activate the nics when used with this board.

Anyone have any ideas?

[laga]

Anyone have any ideas?

That sounds odd. Does it work with a different computer (but using the same NIC + etherboot image)? I assume your etherboot disk matches your NIC.

Have you tried a NIC that has its own PXE stack?

[bodyjarrocks]

Hi all,

Just a report on my experience installing mythbuntu-diskless on Hardy.

1) I had to manually add the atl1.ko NIC driver to the modules file in the CHROOT, so that it would be added to the initramfs. I have an ASUS P5E-VM HDMI.

2) I had to manually add the a kernel modules package to the chroot (via apt-get install in the chroot). This gave me squashfs.ko, which is needed to boot obviously. I would have thought this would be done automatically.

3) I had to change the /etc/exports file for the NFS overlay. The overlay was only exported on 127.0.0.0/24. This was not exposed to the client and it could not boot. I changed it to 192.168.1.0/8

4) After those changes, everything booted up. I noticed that there is no loopback network interface (127.0.0.1 - lo). I don't know if this is a problem.

5) Recorded shows cannot be played back. This probably has nothing to do with diskless, and the issue is being tracked on:
http://www.gossamer-threads.com/lists/mythtv/users/333216#333216
and
http://ubuntuforums.org/showthread.php?t=774470

6) I could start a recording, but I could not stop it from the program guide.

Thanks for all your hard work on this.

P.S. How do stay up to date package-wise with this diskless stuff? Do I need to install the mythbuntu 0.21-fixes repo on my serevr and in the chroot?

Mike

[bodyjarrocks]

SOLVED.

This can be solved by running the following in a terminal:
dpkg-reconfigure tzdata

Set your timezone correctly and the problem is gone. I ran into this problem because I'm using mythbuntu-diskless. The diskless image for the fron end lacked the timezone info but the back end was correct. Since there was a time mismatch, the front end decided that recordings did not exist.

Cheers,
Mike.

[laga]

That's really odd. The ltsp-build-client plugins should have taken care of that. Either they failed or your installation wasn't successful, which I doubt since you have the squashfs image. Very odd.

Maybe you can file a bug at http://bugs.launchpad.net/ubuntu/+source/mythbuntu-diskless and I'll try to reproduce it.

Regarding your previous posting:

1) deserves a bug report, too

2) should be reported, too. You're really having some weird issues :) squashfs.ko lives in linux-ubuntu-modules which should be installed.

3) mythbuntu-control-centre uses the IP address of the master backend to determine the correct network. If that address is still set to 127.0.0.1, the expot won't be usable :) It'd be great if you could report a bug about that as well, but there's probably no good way to fix this.

I'll try to reproduce your problems during the weekend. Thanks for your feedback, it's much appreciated!

[mondo1287]

It seems like fstab is being overwritten each time a diskless client boots. If I change it in the chroot and then rebuild the image, the changes aren't there when a client boots. Also, if I change it from inside the booted client, the changes are gone when the client is rebooted. Am I missing something?

[bodyjarrocks]

Laga, I'll try to raise the bug reports this weekend. Unfortunately, due to WAF (Wife Approval Factor) I can't do any fresh installs :-)

Thanks for you help and thanks laga and all for such a awesome system.

Mike.

[laga]

It seems like fstab is being overwritten each time a diskless client boots. If I change it in the chroot and then rebuild the image, the changes aren't there when a client boots. Also, if I change it from inside the booted client, the changes are gone when the client is rebooted. Am I missing something?


What's the content of /opt/ltsp/i386/etc/lts.conf? Did you create a custom lts.conf somewhere?

[JimSoft]

It seems like fstab is being overwritten each time a diskless client boots. If I change it in the chroot and then rebuild the image, the changes aren't there when a client boots. Also, if I change it from inside the booted client, the changes are gone when the client is rebooted. Am I missing something?


I have exactly the same problem and have spent quite some time trying to figure out why....any clues as I getting desperate now.

CONFIGURE_FSTAB=false is set BTW.

[asmith3006]

I think I'm missing a few points here.

I'm trying to get this up and running (diskless mode) but I can't get my mythbuntu backend to act as a DHCP server. I've checked the dhcpd.conf file and that seems to be ok, I've got "Diskless Server" and "Add DHCP Server" selected and applied in the config interface but I don't get a DHCP address from a client when I connect it (I get a manual one).

Am I missing lots of steps, or is it as simple as checking those boxes?

Thanks for any and all help!

[laga]

but I don't get a DHCP address from a client when I connect it (I get a manual one).



I'm not sure what you're trying to do here. Are you trying to get your client to act as a DHCP server?

[laga]

I have exactly the same problem and have spent quite some time trying to figure out why....any clues as I getting desperate now.

CONFIGURE_FSTAB=false is set BTW.

Odd. I'll try to investigate as soon as time permits. CONFIGURE_FSTAB is handled by /opt/ltsp/i386/etc/init.d/ltsp-client-setup in line 142. Does yours look like this:


configure_fstab() {
if [ -z "$CONFIGURE_FSTAB" ] || boolean_is_true "$CONFIGURE_FSTAB" ; then
echo "/dev/root / unionfs defaults 0 0" > /etc/fstab
echo "tmpfs /tmp tmpfs defaults,nosuid,nodev 0 0" >> /etc/fstab
mount /tmp
fi

[asmith3006]

I'm not sure what you're trying to do here. Are you trying to get your client to act as a DHCP server?I've setup a ubuntu computer as a primary backend server and in the config utility I selected "diskless server" and "dhcp server" but it's not working as a DHCP server.

[laga]

Ah, OK :)

Can you post /etc/ltsp/dhcpd.conf here? Also, is there anything related to DHCP in /var/log/daemon.log?

[duckduckgoose]

Ok so I have tried the following 2 setups:

7.10 server upgraded to 8.04 server overlayed with Mythbuntu 8.04.
&
Fresh Mythbuntu fully patched with proposed.

Both installs fail to make a proper ltsp release, i386.img is not created.

I am not sure what I am doing wrong here, but I see that others have succeeded.

I ran the code:

MCC_DEBUG=true sudo -E /usr/share/mythbuntu-control-centre/bin/mythbuntu-control-centre | tee mcc.log

and will soon have the results to post here.

Anyone else have the same results as me, any ideas what could be going wrong?

Update: The build failed with the following message dialog:

ltsp-build-client failed with exit code 1. maybe you need to have "Allow unsigned packages" checkbox?

however I have installed the latest control-centre .deb from laga
and "Allow unsigned packages" is definitely checked.

Update 2:
I dug into the wiki and came across this:
Download an updated version of the mythbuntu-diskless LTSP plugin:

* $ wget [WWW] http://laga.ath.cx/030-mythbuntu

* and copy it into the right place:
o $ sudo cp 030-mythbuntu /usr/share/ltsp/plugins/ltsp-build-client/Ubuntu/030-mythbuntu


after making this addition, I then built the ltsp image by hand using
sudo ltsp-build-client --mythbuntu --mythbuntu-copy-user-credentials

and it worked, I was able to boot via pxe and access the front end.

I notice Mplayer wasn't installed so I grabbed it via apt, and on reboot it was still there.

next I exported my /var/lib/mythtv/* directories via nfs to the frontend and added them to the frontend /etc/fstab and was able to watch recordings/videos.

however on the next reboot my nfs mount from fstab were gone, so now I just have to figure out how to make my mounts stick across reboots. and I should be good.

-Goose

[laga]

Ok, I need to edit the wiki page or remove the LTSP plugin from my web space. It's outdated and missing lots of stuff.

It'd be interesting to see your mcc.log file (after reinstalling ltsp-server to get rid of the old plugin). I'm sorry that it's not working for you :( but it should be pretty obvious from the mcc.log to see what's going wrong :)

[duckduckgoose]

So I should remove diskless-server role then add it again, run build-client with the command line that logs, and post my mcc.log?

-Goose

[duckduckgoose]

Ok I removed all my kludging and "apt-get --reinstall install" all packages concerned.. ran the control-centre with debugging and the mcc.log shows everything went fine until..

This:

After this operation, 1031MB of additional disk space will be used.
WARNING: The following packages cannot be authenticated!
libavutil1d libavcodec1d libavformat1d libswscale1d ffmpeg libpostproc1d
error: LTSP client installation ended abnormally


"allow unsigned packages" was checked for sure.

So where does that leave us?

-Goose

[laga]

Hi,

what version of the control centre are you running? Use 'apt-cache policy mythbuntu-control-centre' to find out.

[duckduckgoose]

Did some more poking around and it looks to me like the mythbuntu-control-center is not setting $option_accept_unsigned_packages_value


I edited 001-set-accept-unsigned-packages to force it, to see what happens. I know this doesn't solve the problem, but hopefully it will give some clues as to where to look for a permanent fix.

-Goose

Ok I removed all my kludging and "apt-get --reinstall install" all packages concerned.. ran the control-centre with debugging and the mcc.log shows everything went fine until..

This:


"allow unsigned packages" was checked for sure.

So where does that leave us?

-Goose

[duckduckgoose]

Here is the output:

root@media01:/usr/share/ltsp/plugins/ltsp-build-client/Ubuntu# apt-cache policy mythbuntu-control-centre
mythbuntu-control-centre:
Installed: 0.28-0ubuntu1~hardy1
Candidate: 0.28-0ubuntu1~hardy1
Version table:
*** 0.28-0ubuntu1~hardy1 0
500 http://nz.archive.ubuntu.com hardy-proposed/universe Packages
100 /var/lib/dpkg/status
0.27-0ubuntu1 0
500 http://nz.archive.ubuntu.com hardy/universe Packages


-Goose

Hi,

what version of the control centre are you running? Use 'apt-cache policy mythbuntu-control-centre' to find out.

[duckduckgoose]

So I have success after forcing $option_accept_unsigned_packages_value So I must have the wrong mythbuntu-control-centre installed.

Which is the correct version? waiting on your reply.
-Goose

Did some more poking around and it looks to me like the mythbuntu-control-center is not setting $option_accept_unsigned_packages_value


I edited 001-set-accept-unsigned-packages to force it, to see what happens. I know this doesn't solve the problem, but hopefully it will give some clues as to where to look for a permanent fix.

-Goose

[jfix]

Not sure if it is ok to hijack this thread, but as everybody seems to do it ... anyway, here it goes.

I have started to install the 8.04 on a server and a diskless client using pxe and the procedure found at https://help.ubuntu.com/community/MythTV/Install/Hardy/Diskless because the MCC way of doing things didn't work.

Currently, my problem is simply that the time of the backend and the diskless frontend are not the same. BTW, I am in Paris, France. The frontend is 2 hours ahead, like this:

jakob@backend:~$ date
Mon May 12 22:08:16 CEST 2008

jakob@frontend:~$ date
Tue May 13 00:08:34 CEST 2008

Fixing the date manually using date doesn't work as it gets reset on any reboot. Any ideas as to how to fix this permanently?

Thanks for all the good work so far!

--
cheers,
Jakob.

[duckduckgoose]

looking into it, I noticed ntp does not get installed on the image. not sure if there is a reason why but I am going to try to install it and see what happens, I have the same problem.

-Goose


Not sure if it is ok to hijack this thread, but as everybody seems to do it ... anyway, here it goes.

I have started to install the 8.04 on a server and a diskless client using pxe and the procedure found at https://help.ubuntu.com/community/My...Hardy/Diskless because the MCC way of doing things didn't work.

Currently, my problem is simply that the time of the backend and the diskless frontend are not the same. BTW, I am in Paris, France. The frontend is 2 hours ahead, like this:

jakob@backend:~$ date
Mon May 12 22:08:16 CEST 2008

jakob@frontend:~$ date
Tue May 13 00:08:34 CEST 2008

Fixing the date manually using date doesn't work as it gets reset on any reboot. Any ideas as to how to fix this permanently?

Thanks for all the good work so far!

--
cheers,
Jakob.

[laga]

duckduckgoose,

the ~hardy2 version would be appropriate. Unfortunately, it's not in hardy-proposed yet. I messed up some paper work when I tried to get it in and now I'm busy.. I'll try to get it tomorrow. See this thread for the updated deb: http://ubuntuforums.org/showthread.php?t=775489

But I guess you don't need it anymore :)

jfix,

is the hardware clock of your frontend set to UTC or to local time?

[jfix]

duckduckgoose,

the ~hardy2 version would be appropriate. Unfortunately, it's not in hardy-proposed yet. I messed up some paper work when I tried to get it in and now I'm busy.. I'll try to get it tomorrow. See this thread for the updated deb: http://ubuntuforums.org/showthread.php?t=775489

But I guess you don't need it anymore :)

jfix,

is the hardware clock of your frontend set to UTC or to local time?

laga,

not sure at the moment, how can I find out?

[laga]

jfix,

when your computer starts, just enter the BIOS setup (by pressing 'del' or 'F2' - your BIOS should give you a message which lists the correct key when it starts). Then you should be able to see the time. If the time is correct, then it's local time. If it's off by 2 hours, it's GMT ;).

[asmith3006]

Ah, OK :)

Can you post /etc/ltsp/dhcpd.conf here? Also, is there anything related to DHCP in /var/log/daemon.log?

Sorry it took so long to get back (should I start my own thread? I'm not sure what the etiquette is here)



# Default LTSP dhcpd.conf config file.
#

authoritative;

subnet 192.168.0.0 netmask 255.255.255.0 {
range 192.168.0.20 192.168.0.250;
option domain-name "example.com";
option domain-name-servers 192.168.0.1;
option broadcast-address 192.168.0.255;
option routers 192.168.0.1;
# next-server 192.168.0.1;
# get-lease-hostnames true;
option subnet-mask 255.255.255.0;
option root-path "/opt/ltsp/i386";
if substring( option vendor-class-identifier, 0, 9 ) = "PXEClient" {
filename "/ltsp/i386/pxelinux.0";
} else {
filename "/ltsp/i386/nbi.img";
}
}


There's nothing in the log file about the DHCP server. It mentions DHCP, but only client side. To test the system I'm plugging it into it's own switch and rebooting it and then attaching my laptop and checking if I get a DHCP address or not. I'm plugging it back into my internet network so I can get updates and change packages to try and get this working.

Thanks for any and all help.

[asmith3006]

Sorry to do this but *bump*

[laga]

asmith3006,

that config file looks sane.

Let me recapitulate your setup: you've got a server and a client (your laptop). When testing, your (only?) network card in your server is plugged into a switch. Your laptop is also plugged into that switch. Only these two devices are plugged in. Correct?

What's the output of ifconfig -a on the server? What exactly happens if you run sudo dhclient ethX on your laptop to get a DHCP lease? Of course, you'll need to substitute X with the real interface number, ie '0'. Also, what exactly shows in /var/log/daemon.log on the server when requesting a DHCP lease? Is dhcpd running? Check with ps ax | grep dhcpd3

edit: You usually don't have to bump this thread. I try to look at it every day, but sometimes life gets in the way. And no, you don't have to start your own thread :)

[jfix]

laga,

jfix,

when your computer starts, just enter the BIOS setup (by pressing 'del' or 'F2' - your BIOS should give you a message which lists the correct key when it starts). Then you should be able to see the time. If the time is correct, then it's local time. If it's off by 2 hours, it's GMT ;).

ok, it's localtime. I changed it to -2 hours, so that the time is GMT/UTC. now the time is shown correctly in the frontend.
that's how it should be, right? hardware clock set to GMT?

thanks a lot,
jfix.

[asmith3006]

asmith3006,

that config file looks sane.

Let me recapitulate your setup: you've got a server and a client (your laptop). When testing, your (only?) network card in your server is plugged into a switch. Your laptop is also plugged into that switch. Only these two devices are plugged in. Correct?

What's the output of ifconfig -a on the server? What exactly happens if you run sudo dhclient ethX on your laptop to get a DHCP lease? Of course, you'll need to substitute X with the real interface number, ie '0'. Also, what exactly shows in /var/log/daemon.log on the server when requesting a DHCP lease? Is dhcpd running? Check with ps ax | grep dhcpd3

edit: You usually don't have to bump this thread. I try to look at it every day, but sometimes life gets in the way. And no, you don't have to start your own thread :)

The server has two network cards. I eventually want to use it as a router too.

The output of ifconfig is
eth0 Link encap:Ethernet HWaddr 00:90:27:1c:d0:4d
inet6 addr: fe80::290:27ff:fe1c:d04d/64 Scope:Link
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:145 errors:0 dropped:0 overruns:0 frame:0
TX packets:24 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:34009 (33.2 KB) TX bytes:3100 (3.0 KB)

eth1 Link encap:Ethernet HWaddr 00:10:dc:54:74:5e
UP BROADCAST MULTICAST MTU:1500 Metric:1
RX packets:21527 errors:0 dropped:0 overruns:0 frame:0
TX packets:33190 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:27955743 (26.6 MB) TX bytes:23619987 (22.5 MB)

lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
inet6 addr: ::1/128 Scope:Host
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:1978 errors:0 dropped:0 overruns:0 frame:0
TX packets:1978 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:103848 (101.4 KB) TX bytes:103848 (101.4 KB)

The results from ps ax | grep dhcpd3 is 7391 pts/0 R+ 0:00 grep dhcpd3

Thanks for your help on this.

Yes I do have a server (which runs both a back and front end) and then I have a laptop for testing. It's a macbook. I'm testing the DHCP server with the mac booted into mac os. I tried forcing a refresh of DHCP license but no joy.

Seeing as I haven't got very far with this (there's no media files yet) I'm tempted to take out a network card and start from scratch. I was playing around with Webmin and eBox before this and I'm starting to wonder if there is some old installation that's messed up..

Thanks again.

Andrew.

[laga]

Andrew,

unless I'm missing something, neither of your network cards has an IP address. You need to configure the interface which points to your diskless clients first (and verify that /etc/ltsp/dhcp.conf is correct afterwards).

[asmith3006]

Andrew,

unless I'm missing something, neither of your network cards has an IP address. You need to configure the interface which points to your diskless clients first (and verify that /etc/ltsp/dhcp.conf is correct afterwards).
How do I do that? How do I assign one a static address? and how do I know which one is configured for DHCP?

Thanks again.

[ares01590]

I think I found why /etc/fstab is constantly getting written over. The lts.conf is missing "[default]" as the first non-comment line. So getltscfg is erroring out instead of setting the variables.

[blackoper]

I think I found why /etc/fstab is constantly getting written over. The lts.conf is missing "[default]" as the first non-comment line. So getltscfg is erroring out instead of setting the variables.

thanks that helped immensely. Between setting that to default and modifying the file with the standard fstab to include my samba shares by default and then rebuilding the image everything now works automatically

sidequestion:
would upgrading my kernel to 2.6.25 by manually compiling it cause any problems with the diskless booting ability (or package dependencies)? I need the 2.6.25 kernel for port multiplier support (half my drives need port multiplier to work) and no one has been able to get port multipliers to work in 2.6.24 kernel of 8.04 that I can tell. I'm going to test with a clean fedora 9 install to test

[stiev3]

I'm curious as to how to start over fresh with my client images. It seems that my configuration is persisting through the process of deleting and rebuilding img files.

[duckduckgoose]

If you mean that your diskless client specific changes have persisted, that can be fixed by removing the Overlay directory for your diskless clients mac address. The overlay is located at: /var/cache/mythbuntu-diskless/overlay/<YOUR-CLIENTS-MAC-ADDRESS>

I'm curious as to how to start over fresh with my client images. It seems that my configuration is persisting through the process of deleting and rebuilding img files.

Hope this helps,
-Goose

[laga]

sidequestion:
would upgrading my kernel to 2.6.25 by manually compiling it cause any problems with the diskless booting ability (or package dependencies)? I need the 2.6.25 kernel for port multiplier support (half my drives need port multiplier to work) and no one has been able to get port multipliers to work in 2.6.24 kernel of 8.04 that I can tell. I'm going to test with a clean fedora 9 install to test

Mythbuntu-diskless relies on aufs ("another unionfs") which in turn relies on some kernel patches. I think you're better off installing 2.6.25 from Ubuntu intrepid, assuming it's already been uploaded (and is usable). Don't forget to install the corresponding linux-ubuntu-modules package (which contains aufs) and linux-restricted-modules if you need it.

[blackoper]

well I tried following Encom's post on how to add the ATL1 gigabit driver to the boot image but it's not working. I modified his instructions for amd64 image but nothing appears to have changed and I still get the kernel panic after the boot up can't find my network card drive. I have the P5L-MX asus matx board and it has the exact same network card Encom was using lspci id is even exactly the same. Any ideas on what I need to do to get this working? I edited the mac address to be mine which is: 00:1A:92:14:8F:07. I am using mythbuntu 8.04 standard. The only thing different from Encom's setup is the amd64 image. Any ideas?

EDIT: tried i386 image and it didn't work either. Keeps timing out at the detecting eth0 interface section. Basically the driver for the atl1 card isn't loading

the two modified files look like:
modules:
atl1

and for 70-persistent-net-rules:
#add "# PCI device 0x1969:0x1048 (atl1)"
SUBSYSTEM=="net", DRIVERS=="?*", ATTRS{address}=="00:1A:92:14:8F:07", NAME="eth0"


On the server:


>sudo mount -o bind /proc /opt/ltsp/i386/proc/
>sudo chroot /opt/ltsp/i386

>sudo vi /etc/initramfs-tools/modules

add "atl1"

>sudo update-initramfs -u

>sudo vi /etc/udev/rules.d/70-persistent-net.rules

Add the following two lines between the dashes:
-----------

#add "# PCI device 0x1969:0x1048 (atl1)"
SUBSYSTEM=="net", DRIVERS=="?*", ATTRS{address}=="00:1e:8c:7d:59:e8", NAME="eth0"

-------------

Replacing the appropriate MAC address with your own. To find your MAC address look at your DHCP log.

>exit
>sudo umount /opt/ltsp/i386/proc/
>sudo ltsp-update-image

[blackoper]

ok I figured it out. I upgraded packages on the main server when I installed mythbuntu and it broke the atl1 driver. I managed to track it down when I installed mythbuntu on the frontend and it worked fine until I did an update and rebooted.. so moral of story don't update mythbuntu even on a new install :) Now to save config files and reinstall everything.

[laga]

What broke the driver? A kernel update? Is there a bug on Launchpad?

[laga]

Hello,

we finally got a new mythbuntu-control-centre upload which fixes some of the outstanding -diskless issues. Those of you who tested the deb I posted earlier will already have those changes, but it'd be good if everyone interested installed the latest MCC from the hardy-proposed pocket and let me know how it goes.

Here's the Launchpad entry for the source package: https://launchpad.net/ubuntu/hardy/+source/mythbuntu-control-centre/0.28-0ubuntu1~hardy2

Please post your feedback in the following bug report in Launchpad: https://bugs.launchpad.net/ubuntu/+source/mythbuntu-control-centre/+bug/221921

If we get enough positive feedback, the package will be copied into hardy-updates so everyone can get it.

[Verzweifler]

On my Mythbuntu-Server, I have installed some nice perl scripts that - together with mbmon, hddtemp and rrdtool - report the status of my harddisks, memory and the like... It works like a charm.

So, I tried to install those to my Mythbuntu-diskless client as well (in the chroot), but whenever I manually call mbmon or hddtemp, I receive error messages:
perl: warning: Setting locale failed.
perl: warning: Please check that your locale settings:
LANGUAGE = "de_DE",
LC_ALL = "de_DE",
LC_LANG = "de_DE",
LANG = "de_DE"
are supported and installed on your system.

Although the tools finally provide the information requested, the preceeding warnings in the output just trash my scripts... What can I do?

[laga]

I don't know much about these problems, but maybe sudo dpkg-reconfigure locales inside the chroot can help. You probably also need to add de_DE to the list of locales to be generated, but I'm not entirely sure how to do that :)

[jaw1959]

I'm trying to follow the instructions in the "documentation" link, but I seem to have a problem at the step where I "wget http://laga.ath.cx/031-mythbuntu" ... I get a 404: Not Found. Has the file moved, or is the server simply down right now?

[Verzweifler]

I don't know much about these problems, but maybe sudo dpkg-reconfigure locales inside the chroot can help. You probably also need to add de_DE to the list of locales to be generated, but I'm not entirely sure how to do that :)

It is as easy as this:

In the chroot say:
locale-gen de_DE.UTF-8
which will install the requested locale. Afterwards a simple
dpkg-reconfigure locales will register that locale so it appears when you type locale -a.

Make sure to commit your changes afterwards, i.e. afteer returning to the MCC. ;-)

[laga]

I'm trying to follow the instructions in the "documentation" link, but I seem to have a problem at the step where I "wget http://laga.ath.cx/031-mythbuntu" ... I get a 404: Not Found. Has the file moved, or is the server simply down right now?

It's outdated. Everything you need is in the Ubuntu repos now.

[sfalcon1]

How does one update their diskless kernel? I have tried to do it from the client and chroot. I checked the aptitude list and the updated version does not show up, only 2.6.24-16. My Main box is using 2.6.24-19. I have tried to ltsp-update-image, chroot and apt, but nothing is letting me get updated packages. It is stating everything is up to date. Any help is appreciated.

Scott

[WattoDaToydarian]

Hey guys I am having some trouble getting this to work on my server.
My server is running Ubuntu-server 8.04 and I installed mythbuntu-diskless-server i386.
I also have another server running Clarkconnect as my internet gateway/firewall which is running the DHCP service for my network.
My Clarkconnect server is IP 192.168.0.1 and my ubuntu-server is 192.168.0.4.
Clarkconnect uses dnsmasq and this is the configuration that I am using:
# The "interface" parameter is set by the network policy (LAN/DMZ interfaces)
bogus-priv
conf-file=/etc/dnsmasq/dhcp.conf
dhcp-authoritative
dhcp-lease-max=1000

# rootpath option
dhcp-option=17,/opt/ltsp/i386

domain-needed
domain=***********
expand-hosts
no-negcache
strict-order
user=nobody

# define common netboot types
dhcp-vendorclass=etherboot,Etherboot
dhcp-vendorclass=pxe,PXEClient

# Set the boot filename for BOOTP.

dhcp-boot=net:pxe,/ltsp/i386/pxelinux.0,watopia,192.168.0.4
dhcp-boot=net:etherboot,/ltsp/i386/nbi.img,watopia,192.168.0.4
Now what happens when my client box boots is it loads everything correctly until it tries to mount the NBD share.
I see it trying to connect to 192.168.0.1 which does not have my NBD service on it.
Also, on my ubuntu-server I don't see any nbd-server running. Is it suppose to be running or does it somehow start when the client tries to connect?
For the wrong IP problem I have tried adding "nbdroot=192.168.0.4:2000,squashfs" to the default pxelinux.cfg but it had no effect.

Thanks in advance for any help!

[lewbuntu]

I had the same nbd issue during pxe boot, eventually getting dropped into a busybox shell. After fruitless googling effort, I clobbered it to all work over NFS,as I only have one diskless system in use. Nice to have it working, but I'd be very keen to get the overlay right. I'm struggling to find the relevant info on nbd and the overlay.
It'd be a welcome addition to the current community howto and mythbuntu install guide.
Anyone able to shed some additional light on this bit?

[stratusSS]

I am having issues with the diskless front end. I can get both the PXE and the flash drives to boot into the diskless os. Once there I go to MythFrontEnd icon (or whatever its called) and it askes for the back end info... if you put it in wrong it tells you that obviously it cant connect. However the problem is when I put the correct info the screen disappears and I am just staring at the desktop again.

If I take the backend down I go back to the blue screen where it asks me for the information again... I will try and find the log files and post back

[becvert]

I'm having the same problem with nbd.
I'm getting this:
Being: retrying nbd mount...
Error: Connect: Connection refused

nbd-server service is not running and I can't start it
I'm getting this:
** (process:9864): WARNING **: Could not parse config file: Unknown error
** Message: Nothing to do! Bye!
nbd-server.

I made a /etc/nbd-server/config file with a GENERIC section but still won't work.

also I think I have not translated the inet.conf file to the xinet format correctly.

/etc/inet.conf:

tftp dgram udp wait root /usr/sbin/in.tftpd /usr/sbin/in.tftpd -s /var/lib/tftpboot
9571 stream tcp nowait nobody /usr/sbin/tcpd /usr/sbin/ldminfod
9572 stream tcp nowait nobody /usr/sbin/tcpd /usr/sbin/nbdswapd
2000 stream tcp nowait nobody /usr/sbin/tcpd /usr/sbin/nbdrootd /opt/ltsp/images/i386.img


/etc/xinet.d/tftp:
service tftp
{
disable = no
socket_type = dgram
protocol = udp
wait = yes
user = root
server = /usr/sbin/in.tftpd
server_args = -s /var/lib/tftpboot
}

service 9571
{
disable = no
socket_type = stream
protocol = tcp
wait = no
user = nobody
server = /usr/sbin/ldminfod
}

service 9572
{
disable = no
socket_type = stream
protocol = tcp
wait = no
user = nobody
server = /usr/sbin/nbdswapd
}

service 2000
{
disable = no
socket_type = stream
protocol = tcp
wait = no
user = nobody
server = /usr/sbin/nbdrootd
server_arg = /opt/ltsp/images/i386.img
}

some help would be greatly appreciated
thanks

[becvert]

I rebuild my config file with sudo dpkg-reconfigure nbd-server

it contains this:

[generic]
# If you want to run everything as root rather than the nbd user, you
# may either say "root" in the two following lines, or remove them
# altogether. Do not remove the [generic] section, however.
user = nbd
group = nbd

# What follows are export definitions. You may create as much of them as
# you want, but the section header has to be unique.
[export]
exportname = /opt/ltsp/images/i386.img
port = 2000


and now I'm having the Welcome login screen!

[wortmanb]

I just installed 804 as part of upgrading from LinuxMCE, where I did have diskless working though exactly how it worked, I couldn't say.

After going through the directions in the guide, I have a system that's acting as the DHCP server for my network, and which is also supposed to be acting as a diskless server. I built a disk image for i386 (the server is amd64) but when I try to boot a client that used to boot diskless from my lmce server, nothing happens. I see the DHCP server on my new box responding with the lease, but that's as far as it seems to get.

It should be noted that I had to create my own dhcpd.conf file as the one provided basically was a template only, and I have no fine in /etc/nbd-server, where I'd expect to see a config file.

Here's what I currently have defined in my dhcpd.conf file, with all commented lines removed:

ddns-update-style none;
option domain-name "thewortmans.org";
option domain-name-servers 192.168.1.1;
default-lease-time 600;
max-lease-time 7200;
authoritative;
log-facility local7;
subnet 192.168.1.0 netmask 255.255.255.0 {
range dynamic-bootp 192.168.1.210 192.168.1.240;
option broadcast-address 192.168.1.255;
option routers 192.168.1.1;
}

I'm not at all confident that the above is correct. Any pointers, or anything else I should provide? I'm more than happy to work at this as much as it takes to get it working and provide some feedback....

(Whoops, just realized there are a lot more posts here than I'd previously thought. Will try reconfiguring the package and some other things from this thread in the next day or two).

[wortmanb]

I mentioned dhcpd.conf in my previous post, and I'm certain that file isn't correct. I found the template in /usr/share/doc/ltsp-server/examples and modified it and placed it in /etc/ltsp (which directory I had to create manually). How does this relate to the dhcpd.conf file in /etc/dhcp3? What is ltsp and how do I know that it's working properly on my system?

Thanks!

[laga]

I mentioned dhcpd.conf in my previous post, and I'm certain that file isn't correct. I found the template in /usr/share/doc/ltsp-server/examples and modified it and placed it in /etc/ltsp (which directory I had to create manually). How does this relate to the dhcpd.conf file in /etc/dhcp3?


/etc/dhcp3/dhcpd.conf is the default configuration file for dhcp3-server on Ubuntu. However, if /etc/ltsp/dhcpd.conf exists, it is used instead. See the following code from /etc/init.d/dhcp3-server:


# Default config file
CONFIG_FILE=/etc/dhcp3/dhcpd.conf

# Allow ltsp to override
if [ -f /etc/ltsp/dhcpd.conf ]; then
CONFIG_FILE=/etc/ltsp/dhcpd.conf
fi



What is ltsp and how do I know that it's working properly on my system?


LTSP stands for "Linux Terminal Server Project". Mythbuntu-diskless is built on LTSP - it uses ltsp-build-client, for example.

I'm not sure if you're still having problems - maybe you can be a bit more specific on what you did and what doesn't work.

[wortmanb]

...my diskless client is now working!

What I did was:

# dpkg-reconfigure nbd-server
:
# mkdir /etc/ltsp
# cp /usr/share/doc/ltsp-server/examples/dhcpd.conf /etc/ltsp/
# vi /etc/ltsp/dhcpd.conf

And now it's working like a champ! About all I did was change the subnets to match the one I'm using since the addresses (routers and gateways and such) matched up nicely.

[laga]

...my diskless client is now working!

What I did was:

# dpkg-reconfigure nbd-server
:
# mkdir /etc/ltsp
# cp /usr/share/doc/ltsp-server/examples/dhcpd.conf /etc/ltsp/
# vi /etc/ltsp/dhcpd.conf

And now it's working like a champ! About all I did was change the subnets to match the one I'm using since the addresses (routers and gateways and such) matched up nicely.

I'm glad it's working for you now!

How did you set it up before? If you used the control centre, what options did you choose there?

[wortmanb]

I'm glad it's working for you now!

How did you set it up before? If you used the control centre, what options did you choose there?

I believe I followed the directions in the installation guide; I know that I set up the disk image via "Diskless Server" in the MCC app, then added a primary DHCP server to this box as well, disabling the existing DHCP server on my network and giving this one's eth0 and eth1 static addresses (through Applications->System->Network). After all these changes were applied, I rebooted and tried to boot my diskless box. No joy.

I then went looking for the dhcpd.conf file and found the one in /etc/dhcp3, which is where I'm used to finding it. I updated it a few times before discovering and scanning this thread. I performed the steps above and now have a functioning diskless boot setup. I just need to ensure that the frontend works over there and then figure out how to get my ati Remote Wonder USB RF remote working.

Did that cover what you were looking for? I was honestly flying so fast I may have missed some of the early steps. Setting up Mythbuntu has been a dream compared to the last time I set up Myth on a Fedora box, back in 2003 or so!

[laga]

Did that cover what you were looking for? I was honestly flying so fast I may have missed some of the early steps.


Yep, Thanks! I guess I should try to reproduce your problem on a fresh install - if I ever get time to do that ;)


Setting up Mythbuntu has been a dream compared to the last time I set up Myth on a Fedora box, back in 2003 or so!

Cool! I'm glad it's working for you :)

[mal]

I've been fiddling around for a couple of weeks trying to get Mythbuntu-diskless working. My client machine could not get past the tftp loading step and then the kernel loading process. Most of the time I was using the manual installation process from the wiki (https://help.ubuntu.com/community/MythTV/Install/Hardy/Diskless).

I eventually fixed the tftp issue by editing /etc/default/tftpd-hpa to change RUN_DAEMON="no" to RUN_DAEMON="yes". This seems to be needed because there was no tftp entry in /etc/inetd.conf. I don't understand why tftp was not set up to load under inetd.

I then found that the kernel would not load and it appears to have been because I had xinetd installed on my server instead of openbsd-inetd. I had previously installed xinetd for another purpose. When I removed it and reinstalled mythbuntu-diskless-server, I was able to get the system to boot. I gather that there are some configuration options that could get the system to work properly under xinetd, but I took the easy option and just got rid of it.

I now have the system working and am impressed at how fast it loads. This is a great way to run mythtv frontends and eliminates the noise associated with the hard disk in the client.

Mal

[mekis]

Hello,

I have installed the diskless frontend without any problems using an Ubuntu 8.04 server as backend. I did the work on the server loged in as a user that will run the frontend.

My problem is that I don't know how to mount the shares for Videos and Photos. I have added the shares on the server to the /etc/exports file but when I try to configure the frontend using Applications - System - Shared folders and click unlock I get the error:

Could not Authenticate
An unexpected error has occured

Please help

Thanks for a great product, the idea with the diskless frontend is very good.

[mal]

Mekis,

You might get some more expert replies from others, but I found that I had to install nfs-common in the client environment on the server and then edit /etc/fstab on the frontend client machines to include lines to mount the required directories.

Mal

[mekis]

Mekis,

You might get some more expert replies from others, but I found that I had to install nfs-common in the client environment on the server and then edit /etc/fstab on the frontend client machines to include lines to mount the required directories.

Mal

Thank you for the tip mal

I've installed nfs-common and nfs-kernel-server in the client system from the server so these packages are available. After that I've tried to update /etc/fstab on the client with

server://mnt/directory/ /mnt/directory/ nfs auto 0 0

both from the client and from the server but my changes seams to be overwritten. could you please tell me how you did for updating /etc/fstab

[laga]

I have never tried to use NFS shares from a diskless client, so I guess you should listen to mal :) However, to prevent your fstab on the client from getting overwritten, you need to add "[defaults]" to /etc/lts.conf in the chroot. Afterwards, you need to commit your changes (run 'sudo ltsp-update-image' or hit "Commit Changes" in the "diskless" tab in the Control Centre).

Anonymousdog posted a very nice guide to mythbuntu-diskless, btw: http://ubuntuforums.org/showpost.php?p=5464494&postcount=4

[mal]

Mekis,

Actually, I thought I had this working, but I hadn't rebooted the client since I edited /etc/fstab. When I rebooted and checked again, I also found that my amended /etc/fstab was overwritten.

The [defaults] line in /etc/lts.conf did not fix this problem.

As a work around, I added a suitable mount command to /etc/rc.local and this seems to be persistent across reboots. This is probably not a very elegant solution, but mythmusic now works on the client.

Mal

[laga]

The [defaults] line in /etc/lts.conf did not fix this problem.



Where did you add this?

[mal]

Where did you add this?

I added [defaults] as the first non-comment line in /etc/lts.conf in the chroot environment on the server. I then ran sudo ltsp-update-image on the server (outside the chroot).

Note that the /var/cache/mythbuntu-diskless/overlay/.../ directory still has the original fstab file, but the rc.local file is my amended version.

Mal

[mekis]

Thank you Laga for pointing me to this guide:



Anonymousdog posted a very nice guide to mythbuntu-diskless, btw: http://ubuntuforums.org/showpost.php?p=5464494&postcount=4

I followed it and now everything works, including the updated /etc/fstab. I found it specially valuable that Anonymusdog stressed the use of MCC for doing as much as possible. The more I start to understand of the system the more impressed I am of it's functionality.

[lewbuntu]

This took me some time and frustration to work around persistent nbd path issues, so hopefully this info will save someone else the pain.
Those attempting to set up using a separate dnsmasq server to the mythbuntu diskless server, the following may help.

When using openwrt for your dnsmasq server, the traditional dnsmaq.conf addition of ...
dhcp-option=17,192.168.10.xx:/opt/ltsp/images/amd64.bin

... or derivatives and permutations, of the above may find it does not work.
In my instance the nbd requests were continually made to the dnsmasq server, and no doubt failing.
I had to use the following work around...

dhcp-host=00:aa:d0:55:ab:cd,net:dc,192.168.100.22 # mac address of diskless client - name & IP
dhcp-boot=net:dc,os,mythtv-bkend,192.168.100.10
dhcp-option=dc,17,"192.168.100.10:/opt/ltsp/images/amd64.img"

where dc is name of diskless client, mythtv-bkend name of diskless server

I'm sure there is a cleaner more eloquent way to achieve this, but the above works for me.

BTW my nbd-server path is also exported as /opt/ltsp/images/amd64.img

Hope it helps.


Lew

[laga]

lewbuntu,

I have a similar setup. Here's what works for me in dnsmasq.conf:


dhcp-boot=ltsp/i386/pxelinux.0,prometheus,192.168.0.137


prometheus is the host name of the box at 192.168.0.137.

[lewbuntu]

Hi Laga,

I had a similar entry ...
dhcp-boot=net:pxe,ltsp/amd64/pxelinux.0,mythtv-bkend,192.168.0.x

This works fine for the pxe boot, but I had not trouble getting the tftp stuff to work, it was the nbd mount point that would fail.

As far as I could determine, the pxe image would by default attempt to connect to the nbd using the IP of the dhnsmasq server;unless I specified the rootpath and file via option 17.

It seems the "net:pxe" addition is what caused my issues. As you say a simple setting of dhcp-boot=ltsp/i386/pxelinux.0,prometheus,192.168.0.137 does indeed work perfectly!
But my point is, if you want to do it a really really hard way then ...

Prometheus stole fire from the Gods as I recall.

The ego's a bit singed, but what can I say?

Thanks Laga,

Lew

[Disparu]

I might join the diskless squad if you guys approve of this frontend.

http://users.upwardaccess.com/plong/images/gx620_link_2.jpg

has a well support chipset, lga 775 so there is potential for hd playback. insanely small, and has dvi out.

[anonymousdog]

Does it not work?

[gmoser]

I have been working on this non stop for the past 7 or 8 hours... and although it's been rather fun and a learning experience I would really like to be done with it...

I have a WRT-54G router. I have DNSmasq set as:
dhcp-boot=opt/tsp/i386/boot/pxelinux.0,myth-backend,192.168.69.69

I've tried everything I could find from google, but no luck. The file location on my backend is opt/ltsp/i386/boot/pxelinux.0

I know it is getting IP, and it's hitting the back end (when I reboot and the backend is off it takes longer and replies with ARP failures instead of PXE-E3B "TFTP Error - File not found"

ANyone have experience on DD-WRT DNSmasq configs?

[Dilligaf]

I assume that opt/tsp/ in your post is a typo and is really opt/ltsp. This is what I use:

next-server 192.168.1.99;
filename "pxelinux.0";
option root-path "/home/pxe";

[anonymousdog]

http://ubuntuforums.org/showthread.php?t=871867

[laga]

I've tried everything I could find from google, but no luck. The file location on my backend is opt/ltsp/i386/boot/pxelinux.0



It should live in /var/lib/tftpboot/ltsp/i386/.

Usually, tftpd is run chrooted, so it can't access /opt/. Make sure that /var/lib/tftpboot/ltsp/i386/ exists (run 'sudo ltsp-update-kernels') and use "ltsp/i386/pxelinux.0" as the boot path.

[gmoser]

It should live in /var/lib/tftpboot/ltsp/i386/.

Usually, tftpd is run chrooted, so it can't access /opt/. Make sure that /var/lib/tftpboot/ltsp/i386/ exists (run 'sudo ltsp-update-kernels') and use "ltsp/i386/pxelinux.0" as the boot path.

I got in my DNSmasq:
dhcp-boot=ltsp/i386/pxelinux.0,myth-backend,192.168.69.69

But now all I get is PXE-T02: Only absolute filenames allowed
AND PXE-E3C: TFTP Error - Access violation.

I ran sudo ltsp-update-kernels and it did it, but I can't seem to get in.

I built the files from the MCC and read in the past it was buggy, but I was assuming that the bugs were resolved. Should I rebuild the files another way?

[laga]

I got in my DNSmasq:
dhcp-boot=ltsp/i386/pxelinux.0,myth-backend,192.168.69.69


Hum. Can you prepending a / to ltsp/i386/pxelinux.0?



I built the files from the MCC and read in the past it was buggy, but I was assuming that the bugs were resolved. Should I rebuild the files another way?

If /var/lib/tftpboot/ltsp/ exists and looks good, then MCC shouldn't be a problem at this stage.

[gmoser]

I do that:
dhcp-boot=/ltsp/i386/pxelinux.0,myth-backend,192.168.69.69

And now I get: PXE-T02: Forbidden directory.

I think I have user rights issues on my tftp folders. I modified a file (I think INETD something or another) from another suggestion and perhaps this is what is causing this.

Is there a terminal command I can reset the user rights for tftp?

[gmoser]

I also tried the MCC maintenance options and connect to client eth0 and commit changes...

I get a bunch of changes happen and then at the bottom:

Info: port 2000 is already defined wit /opt/ltsp/images/i386.img in inetd.conf
Info: taking no action

Is that part of the problem?

[shawndw]

um i installed the mythbuntu-diskless-server on my 8.04 box hoping that i could get mythbuntu to boot diskless however I decided to uninstall it and now I'm left with a half broken ltsp install and can't get tftpboot working again even after reinstalling it and the DHCP3-server. what exactually did the package leave behind so i can go back and clean it up

[sgtstadanko]

try:

sudo apt-get autoremove mythbuntu-diskless-server

that will grab the stuff installed along with it.

[nuneza]

Hello All,

I am trying to set up diskless frontends on 8.04 and followed the MCC process. I get all of the waty through the process but my front end's wont boot over the network. I do a sudo netstat -apn | grep ":69 " to see if the tftpd-hpa process is running but indicated that it is not. My inetd.conf has the line
2000 stream tcp nowait nobody /usr/sbin/tcpd /usr/sbin/nbdrootd /opt/ltsp/images/i386.img but the following message appears in the daemon.log:
Oct 31 16:06:43 zeus inetd[5646]: 2000/tcp: bind: Address already in use.

My /opt/ltsp/i386 directory is populated as is my /var/lib/tftpboot/ltsp/i386 directory.

My frontend gets a dhcp address but stops after a TFTP open timeout message x3.

I would certainly appreciate any help tracking down what's happening.

Thanks.

[nuneza]

duh. Answer found here: http://ubuntuforums.org/showthread.php?t=148027&highlight=tftpd+start

Sorry for grapping your (mental) processor cycles unnecessarly!

[BjBlaster]

Is there a howto on getting this to work? I have several EPIA M10000 frontends around the house that are diskless and can't seem to get them working with Mythbuntu. I used to use Knoppmyth and that just worked, so I gather there is more to it than just enabling the DHCP sever and downloading that huge image and pressing apply. It finally builds an image but I can't get the DHCP server to give the clients an ip address or image for that matter....

Any good howtos or posts I should follow? I'd like to use mythbuntu instead of knoppmyth because It works out of the box on my backend/frontend machine no worries at all with my USB tuners so I would like to make the switch :)

Cheers

Bj

[anonymousdog]

I think your biggest issue now is DHCP not working. What is the output of 'grep dhcpd /var/log/syslog'? Are any dhcp clients working? Any chance you're running a firewall on your backend?

[BjBlaster]

I will check...

[nuneza]

Hi,

First a tip for all to use...to use XvMC from a diskless client the XvMCConfig file in the overlay directory has to be changed to "libXvMCNVIDIA_dynamic.so.1". f you dont do this, you call tell it to use XvMC but it actually uses ffmpeg.

Now to my question.
I can't get names resolution working for my diskless clients. The diskless client can resolve external names fine via dns and also internal names (I do reverse dns lookup internally). The problem is that I can't resolve the clients address by name.

I enabled wins (installed winbind and edited nsswitch.conf, and got the same behavior. When I look at my dhcp server and see that the client is receiving an IP address, the hostname is blank.

I tried creating the dhcp3 directory and dhclient.conf files in the overlay directory so I could specify the hostname to be sent.

I would really apreciate any advice on how to get this working.

Thanks.

[rmessner]

I seem to be having a problem with this which I have found referenced on the web but not fixed. First off I am using mythbuntu 8.1 as the backend server. When attempting to boot the PXE client I get the kernel and initrd loaded fine. What happens is the machine hangs on the line where it brings up the ethernet interface. I have tried using multiple machines and a virtual machine in vmware, all have the exact same problem. I have tried everything I can come up with to find the problem with no apparent solution. The only thing not standard about my config is I'm using a Tomato firmware based firewall for the dhcp server. My custom dnsmasq config is just dhcp-boot=ltsp/i386/pxelinux.0,name,ip which gets the kernel and initrd with no problem. Any help would be most appreciated.

[laga]

I tried creating the dhcp3 directory and dhclient.conf files in the overlay directory so I could specify the hostname to be sent.


The name of the client is its MAC address without colons. It can't be changed without some manual fiddiling. By default, it uses the host name to find the overlay directory so you'll lose your settings if you're not careful. /etc/initramfs-tools/conf.d/ltsp in the chroot should have everything you need.. you can also specify a lot of thing as boot options.

[laga]

What happens is the machine hangs on the line where it brings up the ethernet interface.

What line is this exactly?

[pharnet]

How feasible is to run LTSP thin client and Mythbuntu Diskless on the same machine. Does Mythbuntu Diskless mind clients built with a different chroot (fati386 in this case)?

I know this is way outside the normal install but I have been working on it for several hours now. I have hit a bit of a wall. I think I am close. The diskless client loads the kernel and loads a system with only a cli login (no gui). When I login I seem to be in the overlay. One item I notice is that while starting up, one of the last few items before the cli logon is "mounting local filesystem [fail]".

The thin client image is i386 and the mythbuntu diskless image is fati386. The thin client is working properly. It is an Ubuntu 8.10 64bit workstation with Mythbuntu packages installed. I am targeting i386 for the thin and diskless clients.

Is this a goofy idea or is it feasible?

[pharnet]

Rebuilt the image. It worked. The mounting local filesystem [fail] message was still there but it didn't seem to mind. Now I will try to add just the perfect extras to my image and then I will forget about it until the next Ubuntu revision.

I threw it out there because I had a vague memory about someone saying that the diskless image didn't like to be renamed.

It still may be a goofy idea but it seems to be feasible.:)

[patrickdk]

I have played with this for the last few days and found two issues.

First was setting up bluetooth on diskless frontends. The normal script is call S25bluetooth and the overlay mount is also S25. This causes bluetooth init to not see already paired devices, till bluetooth is restarted. I personally moved it to S29bluetooth and now I have no issues on rebooting.

The second issue is more annoying, and is when using autofs. I use the auto.net to mount the video, music, and pictures shares. The issue is there is no /net dir. Adding this /net to the compressed image doesn't help, it still acts like /net is missing. But if the /net dir is created on the overlay, everything works as expected (once autofs is restarted). I don't see any easy fix for this.

[patrickdk]

Nope, seems autofs wasn't too bad to fix either.

Adding /net to the disk image, and changing S19autofs to S28autofs fixed it up.

[pharnet]

Running thin client and Mythbuntu diskless does appear to be an issue when you run /usr/sbin/ltsp-update-image. To me it appears the script uses the --arch option for the resulting image name and does not have a --chroot option. I modified the script to use fati386 as the image name and specified the proper port for the nbd-server and was successful.

Seems pretty quiet here, have the 8.10 testers moved on somewhere else?

Once again thanks Laga et al for all the work.

[ttabbal]

I've been trying to get this setup running. I followed the wiki instructions, but the wget step fails. I checked with a browser and the file does not exist on that server. Has it been included in an update? I can build everything without it and boot the client. The client boots the kernel then stops without much info. It does have a message from the network driver about the connection being down when it initializes. That's about it. It's like the kernel starts up, then init doesn't take over.

With the splash screen, I get the Mythbuntu logo then a black screen and nothing else happens. Is there anything I can do to debug this further? I would love to go diskless in the frontends.


EDIT:

I decided to try the GUI. MCC detected the image, so I deleted it and had MCC create a new one. The new one boots in VMWare. I'm remote, but I'll test a hardware boot tonight. It would be nice if CLI were supported, as this server is headless. But VNC worked, so I can do it that way if I need to. I could probably get X working over the network as well if I needed to.

[ttabbal]

OK. The client boots. I can get the GUI and such. However, my hardware is too recent for the installed kernel and NVidia driver. If I update the kernel on the server, does that update the client image as well? Is there a binary for the latest kernel or do I need to go get the source and do it myself?

Server is running Ubuntu Server 8.04. I don't mind updating the kernel and such but I'd prefer to avoid dist-upgrade for now.

UPDATE:

I can't get the NVidia driver to install. Either from packages or the installer from NVidia. It claims to install, but X can't seem to use it. The screen flashes a few times, then X comes up in VGA mode. It's also unusable, something is off as I can't see the whole screen. The GPU is an onboard GF8200 on an Asus M3N78-VM motherboard. Does the driver need to be installed in the base image? I've been doing it on the client machine so I can just delete the overlay directory.

And is it possible to not use squashFS? The rebuild time is mildly irritating when I do updates.

[lee3521]

I would like to test a Mythbuntu-diskless workstation with the client having a wireless NIC card. Is this possible? I am planning on using a flash drive as a loader.
Anyone have any hints for me?

[laga]

I would like to test a Mythbuntu-diskless workstation with the client having a wireless NIC card. Is this possible? I am planning on using a flash drive as a loader.
Anyone have any hints for me?

You will need to write a script which brings up your wireless NIC before a connection to NBD is made. That's not too hard if you know a thing or two about shell scripts and linux. You will either have to modify /etc/initramfs-tools/scripts/mythbuntu_nbd in the chroot or add a new script which is run before the root file system (nbd) is mounted.

Let me know how it goes, I'd love to see your script.

[ttabbal]

I would like to test a Mythbuntu-diskless workstation with the client having a wireless NIC card. Is this possible? I am planning on using a flash drive as a loader.
Anyone have any hints for me?

It would probably work if your flash drive has the necessary kernel driver for the wireless NIC. It would be very slow though. With flash sticks costing <$20 for 8GB, I don't see any reason to mess with it. Just install to the flash stick and call it a day. I finally gave up on diskless because of the NVidia driver issue I posted about earlier. The flash drive was up and running in about 30 min. The only issue I had with the flash drive was the amazingly bright activity LED. A little black electrical tape fixed that right up though.

[peonuser]

I am using 8.1 mythbuntu. Di you need testers for it?

[Pastinakel]

May I add one (very shamefull) comment: I spent the last 3 days trying to figure out how on earth one could possibly make that stupid Mythbuntu box do his thing: building a diskless image. Different versions, different archs, fresh installs, manual building, I tried it all. The MCC kept (very) quiet each and every time I pressed that "Build Image" button...

... I just found out you have to press "Apply" after hitting the "Build Image" button.

Yes, yes, academic education and all, I know...:(:(:(


One more thing... After I had the client booting, I found it stopped somewhere halfway, completely locked. The cause was the nfs share not being correctly exported, resulting in the NFS server refusing the connection from the client. I changed
/var/cache/mythbuntu-diskless/overlay/ 127.0.0.0/24(rw,no_root_squash,async,no_subtree_check)
to
/var/cache/mythbuntu-diskless/overlay 10.0.0.0/24(rw,no_root_squash,async,no_subtree_check)
in /etc/exports

I have no idea who put that silly netmask (127.0.0.0/24) there. I think it is a completely nonsense netmask for a NFS share.

[Verzweifler]

I consider to switch my backend from 32bit to 64bit, but I am not sure whether it will still be able to manage/build/provide diskless images to my 32bit frontend.
Is it possible to build 32bit mythbuntu-diskless images on an AMD64 server system? If not, I would better stick to a 32bit installation of mythbuntu 9.04...

[nuneza]

Hi all,

I've been running Mythbuntu 8.04 with 5 diskless clients for a year or so and it works great. I aquired some new motherboards with onboard Nvidia 8200 GPUs but the version of the nvidia driver that the diskless installation uses (169) doesnt support this chipset. I need to upgrade to a later/ latest nvidia driver but I can't figure out how to do it in/for the diskless environment.

Can anyone point me in the right direction? I noticed that when I tried 9.04, there were several version of the restricted drivers available. Does anyone know what package I can install to do the same in 8.04? I ended up not using 9.04 because there were some problems with the diskless setup that made it not workable for me.

Thanks.

[ttabbal]

Hi all,

I've been running Mythbuntu 8.04 with 5 diskless clients for a year or so and it works great. I aquired some new motherboards with onboard Nvidia 8200 GPUs but the version of the nvidia driver that the diskless installation uses (169) doesnt support this chipset. I need to upgrade to a later/ latest nvidia driver but I can't figure out how to do it in/for the diskless environment.

Can anyone point me in the right direction? I noticed that when I tried 9.04, there were several version of the restricted drivers available. Does anyone know what package I can install to do the same in 8.04? I ended up not using 9.04 because there were some problems with the diskless setup that made it not workable for me.

Thanks.


Interesting that 9.04 doesn't work for you. It worked better for me. For the drivers, that's one thing I never did get working right on 8.04. If the package is available, just chroot into the directory for your diskless client install. The default is /opt/ltsp/<arch> so mine is /opt/ltsp/amd64. Once there, just use apt-get or aptitude as usual to install the packages. "exit" from the command line to get back to the main system prompt and do "ltsp-update-image" to rebuild the compressed disk image your clients use.

Is it the chipset or the onboard GPU you need a driver for? The chipset is handled by the kernel. The GPU is probably best installed with the "nvidia-glx-180" package. I use the JYA repository for that as I use VDPAU and need the patched versions he maintains there. I think the default repo has the 180 series drivers though.

[ttabbal]

I consider to switch my backend from 32bit to 64bit, but I am not sure whether it will still be able to manage/build/provide diskless images to my 32bit frontend.
Is it possible to build 32bit mythbuntu-diskless images on an AMD64 server system? If not, I would better stick to a 32bit installation of mythbuntu 9.04...

There is a command line option to select the target arch for the client image. I'm sure the GUI tool probably has it as well. For "ltsp-build-client" you want "--arch i386". Can't help on the GUI tool as I don't use it here.

[nuneza]

Thanks for the response. 9.04 diskless MCC would not let me complete the configuration of the client environment. It would lock up during some of the steps. I found lots of posts of people having the same problem so I just abandoned the effort until I get a chance to try it again.

I have tried to install the nvidia driver from their site in the chroot environment but it errors out with a complaint that it can't find something it requred. (I'm not at home now so I can't try it) I assumed it was because the chroot is not an actual computer so for example, it wont see an actual GPU. I'll try it again and capture the specific error.

It is the GPU driver that I need. The chipset seems fully supported, at least NIC and sound work fine.

I'll try the 180 package. From what I can tell the 8200 chipset is supported in that version.

Thanks again for the reply.

[WattoDaToydarian]

I just installed ubuntu-server 9.04 on my server and then followed the instructions here > MythTVInstallHardyDiskless (https://help.ubuntu.com/community/MythTV/Install/Hardy/Diskless)
I have it booting and loading X at first but the second time I boot, X won't start.
I am also seeing a TON of SQUASHFS error messages in the syslog (see attached).
I think there is an nfs overlay issue because I see it mounted twice to the same place.
Here is my mount output:
rootfs on / type rootfs (rw)
none on /sys type sysfs (rw,nosuid,nodev,noexec)
none on /proc type proc (rw,nosuid,nodev,noexec)
udev on /dev type tmpfs (rw,mode=755)
/dev/nbd0 on /rofs type squashfs (ro)
192.168.0.1:/var/cache/mythbuntu-diskless/overlay/ on /cow type nfs (rw,vers=3,rsize=524288,wsize=524288,namlen=255,ha rd,nointr,nolock,proto=tcp,timeo=7,retrans=10,sec= sys,addr=192.168.0.1)
aufsroot on / type aufs (rw,si=be806e4d,xino=/tmp/.aufs.xino,br:/cow/0022157669f6=rw:/rofs=ro)
192.168.0.1:/var/cache/mythbuntu-diskless/overlay/ on /cow type nfs (rw,vers=3,rsize=524288,wsize=524288,namlen=255,ha rd,nointr,nolock,proto=tcp,timeo=7,retrans=10,sec= sys,addr=192.168.0.1)
tmpfs on /lib/init/rw type tmpfs (rw,nosuid,mode=755)
fusectl on /sys/fs/fuse/connections type fusectl (rw)
varrun on /var/run type tmpfs (rw,nosuid,mode=755)
varlock on /var/lock type tmpfs (rw,nosuid,nodev,noexec)
tmpfs on /dev/shm type tmpfs (rw,nosuid,nodev)
devpts on /dev/pts type devpts (rw,nosuid,noexec,gid=5,mode=620)
tmpfs on /tmp type tmpfs (rw,nosuid,nodev)

Anyone have a clue:confused: