LinHES Forums
http://forums.linhes.org/

Diskless Frontend - NFSroot
http://forums.linhes.org/viewtopic.php?f=11&t=10098		 Page 16 of 17
Author:  djb61230 [ Sun Nov 08, 2009 8:36 am ]
Post subject: 
Greg Frost wrote:
Out of interest, how long has your backend been up? The script works by checking /var/log/daemon.log and I just booted one of my frontends and found that the mount logs went to /var/log/daemon.log.1
So it looks like the logs have been rotated, but its still writing to the old log. If this is the case with your system, rebooting the backend may solve your problem (because the logs will go to /var/log/daemon.log again).


Yeah I noticed that yesterday as I added a frontend to our "exercise" room. I just hacked on the config_diskless_frontend.bash to look in the .1 file instead of rebooting. :)
Author:  iscraigh [ Sun Nov 08, 2009 12:40 pm ]
Post subject: 
Unfortunately that is not my issue. The ip is in the log I rebooted started frontend and checked again.

Craig
Author:  Greg Frost [ Sun Nov 08, 2009 2:17 pm ]
Post subject: 
Can you pastebin the log or make it available somewhere for download?
Author:  yaplej [ Fri Dec 11, 2009 1:03 am ]
Post subject: 
Can the DHCP portition of this be disabled like in R5? I recall in R5 I had to comment out "allow booting", and the network router could issue IP address while the KM box would reply to bootp requests.

When I started to run the R6 config_diskless_frontend.bash script it was asking me for a range of IP addresses to assign to the frontends. I cancled it because I dont want it doing DHCP. Can this be disabled in this newer script too?

Thanks.
Author:  yaplej [ Tue Jan 05, 2010 1:02 am ]
Post subject: 
Hey all,

I am getting a BE, and diskless FE setup. I have the FE booting after having to fix some video driver problems, but now when I rerun config_diskless_frontend.bash lock the MAC address it says that its unable to automatcially determine the MAC address.

Is there a way to manually specify the MAC address so I can start setting up another diskless FE?

Thanks.
Author:  Greg Frost [ Tue Jan 05, 2010 4:03 am ]
Post subject: 
The script is looking for a line in /var/log/daemon.log that looks something like this:
Code:
Jan  4 22:39:55 mythtv mountd[4461]: authenticated mount request from 10.0.0.21:769 for /nfsroot/fedell (/nfsroot/fedell)

from this it gets the IP address that was served.

It then looks for a line like:
Code:
Jan  4 22:39:55 mythtv dhcpd: DHCPACK on 10.0.0.21 to 00:06:5b:7c:8e:5e via eth0
to determine the MAC address. You could put those details in the file so it can find them.
Author:  yaplej [ Tue Jan 05, 2010 3:04 pm ]
Post subject: 
Thanks Greg,

I was able to save the config by renaming the /tftpboot/pxelinux.cfg/default file to 01-00-14-85-11-18-87 this will be the boot file for that particular MAC address. The extra 01- in front means its the ARP type 1 address.
http://syslinux.zytor.com/wiki/index.php/PXELINUX

So this can be used so the DHCP server does not require IP addresses, and the router, or whatever device you have running can provide IP addresses even for PXE booting.

I ran into another problem when trying to run the script again to setup my 2nd FE though. It seems like the script does not like the "-" in the config file name. Maybe that could get fixed?

config_diskless_frontend.bash: line 65: 0x0-: syntax error: operand expected (error token is "-")

Edit: As a workaround I just moved the config file up one directly level, and reran the script to create a new "default" config file. I then booted my 2nd FE, and moved the 1st FE config file back into the pxelinux.cfg directory. After I got the MAC address of my 2nd FE I renamed the new "default" config file to 01-2nd-FE-MAC-Address, and all is well.
Author:  Greg Frost [ Wed Jan 13, 2010 6:15 am ]
Post subject: 
I think the reason the script does not map the mac address properly sometimes is a logrotation issue. See this flyspray task:
http://linhes.org/flyspray/index.php?do ... ask_id=611
Author:  RacerX [ Sun Mar 14, 2010 7:36 pm ]
Post subject: 
Hello

Since I migrated to R6.02 last week I had to redo all of my diskless workstations. Right now I'm able to create a custom image as usual.
It works fine but I have two issues and was wondering if others have a similar issue.

Here is the log from my diskless frontend.

The first issue is that I'm unable to permanently change the theme. It is always Metallurgy Widescreen. Moreover, while I can change this theme to Blootube for my Diskless frontend that's 1024x768 Standard TV. When I change the theme as soon as I exit out of the frontend it rolls the theme back to Metallurgy next time I launch the frontend. It's like I do not have permission to change it permanently. I tested it also launching the frontend as root and it also rolls back the change. Additionally why does it show 1024x768 and then pick a theme that is 1280x720? Furthermore, how does it pick Metallurgy Widescreen Theme when I need a standard width theme like Blootube etc?

Code:
2010-03-14 19:08:54.331 Using screen 0, 1024x768 at 0,0
2010-03-14 19:08:54.385 MythUI Image Cache size set to 20971520 bytes
2010-03-14 19:08:54.386 Enabled verbose msgs:  important general
Xlib:  extension "Generic Event Extension" missing on display ":0.0".
2010-03-14 19:08:54.405 Primary screen: 0.
2010-03-14 19:08:54.406 Using screen 0, 1024x768 at 0,0
2010-03-14 19:08:54.408 Using theme base resolution of 1280x720
2010-03-14 19:08:54.418 LIRC, Error: Failed to connect to Unix socket '/var/run/lirc/lircd'
         eno: No such file or directory (2)
2010-03-14 19:08:54.419 JoystickMenuThread Error: Joystick disabled - Failed to read /home/mythtv/.mythtv/joystickmenurc
2010-03-14 19:08:54.681 Using the Qt painter
2010-03-14 19:08:54.692 Removing stale cache dir: /home/mythtv/.mythtv/themecache/blootube.1024.768
2010-03-14 19:08:54.847 Loaded base theme from /usr/share/mythtv/themes/metallurgy/base.xml
2010-03-14 19:08:55.025 Loaded base theme from /usr/share/mythtv/themes/default-wide/base.xml
2010-03-14 19:08:55.080 Loaded base theme from /usr/share/mythtv/themes/default/base.xml
2010-03-14 19:08:55.097 Current MythTV Schema Version (DBSchemaVer): 1244
Xlib:  extension "Generic Event Extension" missing on display ":0.0".
Xlib:  extension "Generic Event Extension" missing on display ":0.0".
2010-03-14 19:08:55.699 Desktop video mode: 1024x768 60.006 Hz
Xlib:  extension "Generic Event Extension" missing on display ":0.0".
Xlib:  extension "Generic Event Extension" missing on display ":0.0".



The second issue is the path for Lirc. It's calling /var/run/lirc/lircd

Code:
2010-03-14 19:08:54.418 LIRC, Error: Failed to connect to Unix socket '/var/run/lirc/lircd'
         eno: No such file or directory (2)


shouldn't it be?

/nfsroot/Bedroom/var/run/lirc/licd

Any help, comments would be great
Author:  Greg Frost [ Sun Mar 14, 2010 8:15 pm ]
Post subject: 
From the FEs perspective, it doesnt see /nfsroot/Bedroom/... but its root is actually the contents of that directory. Is the lircd deamon running? I think that error is saying something like "the socket that the lircd deamon uses to send button presses to applications is not present" which would most commonly occur if lircd was not running for some reason.

As for the first prob, I have seen occasions where a host specific setting has been set in the database as a global setting (hostname of null).

Im not home, so im guessing about the syntax, but if you do something like:

Code:
mysqldump --skip-opt mythconverg settings | grep "name of the setting"

where "name of the setting" is the name of the theme setting, and you see an entry without a hostname, that could be your problem.
Author:  RacerX [ Sun Mar 14, 2010 9:24 pm ]
Post subject: 
Greg

I ran the command
mysqldump --skip-opt mythconverg settings | grep "metallurgy"

Code:
INSERT INTO 'settings' VALUES ('OSDTheme', 'metallurgy','Livingroom');
INSERT INTO 'settings' VALUES ('Theme', 'metallurgy', 'Bedroom';

I was reading your other post about the ACER Atom PC and I was thinking about getting one if they worked out so I moved my Livingroom PC up to my Bedroom since it's faster. That created a problem I see because right now I do not have a PC in the Living Room. So how do I get it out of the database?

Also I enabled the snapstream firefly on my Master BE/FE and then rebooted my Bedroom PC but it did not change the error.

Thanks much again for the help :wink:
Author:  Greg Frost [ Mon Mar 15, 2010 4:43 am ]
Post subject: 
what is the theme setting for all hosts?
Code:
mysqldump --skip-opt mythconverg settings | grep Theme

With regard to your remote, you should enter the service menu *on the frontend* not the backend and select the remote for the frontend.

To get rid of settings yo no longer need:
Code:
mysql mythconverg -e "delete from settings where hostname='fename'"


The Revo has worked out brilliantly as a FE, but Im still transitioning to my second revo to replace my backend. I bought a Welland NAS to stick my 1Tb drive from my backend and it is going to be a slow process copying my recordings onto it.
Author:  RacerX [ Mon Mar 15, 2010 3:08 pm ]
Post subject: 
Greg

Well I sorted both problems out today with this command

Code:
mysql mythconverg -e "delete from settings where hostname='fename'"


Looking at it again since I'm a long time user of the Diskless script I had all the pc's that I've ever setup in there, quite messy. In other words time to do some housekeeping!!! Basically I removed all of them except my Master Be/FE and one that I did not recognize. Then I rebooted my Bedroom frontend and it told me to enter my default language. Now I'm back to normal! I changed it to Blootube and it saves it properly. Better still my Firefly remote now works properly!!!

Outstanding :wink:
Author:  phisherpryce [ Thu Aug 19, 2010 12:44 am ]
Post subject: 
I setup a Linhes 6.03 backend and followed instructions to install the diskless-legacy package and run the config script. I commented out the 'allow bootp' as I already have another device acting as a dhcp server.

I'm able to network boot my frontend but I get a couple NFS related failures and a video driver issue:

Quote:
fail: nfslock: unable to change to the service directory: file does not exist

No state is present for card NVidia
Unknown hardware: "HDA-Intel" "Generic 10de NVIDIA MCP7A HDMI" "HDA:10ec0662,10438376,00100101 HDA: 10de0007,10de0101,00100100" "0x1043" "0x8376"
Hardware is initialized using a guess method
No state is present for card NVidia

::Mounting Network Filesystems
mount.nfs: /data/var/cache/pacman is busy or already mounted


The bootup process stops at 'Parent daemon running'. I'm not sure if the nfs errors are causing the problem or the video error.

Any ideas? Thanks.
Author:  poflynn [ Sun Aug 22, 2010 6:45 pm ]
Post subject: 
phisherpryce wrote:
I'm able to network boot my frontend but I get a couple NFS related failures and a video driver issue:

Quote:
fail: nfslock: unable to change to the service directory: file does not exist

No state is present for card NVidia
Unknown hardware: "HDA-Intel" "Generic 10de NVIDIA MCP7A HDMI" "HDA:10ec0662,10438376,00100101 HDA: 10de0007,10de0101,00100100" "0x1043" "0x8376"
Hardware is initialized using a guess method
No state is present for card NVidia

::Mounting Network Filesystems
mount.nfs: /data/var/cache/pacman is busy or already mounted


The machine you built out your mbe on had an nVidia GFX card, this client machine has an intel-based GFX card (I think, please let us know. It may have nVidia on the client too & if so I suspect it is a different nVidia chipset that is not supported by the nVidia driver you have installed currently). You need to manually uninstall the current nVidia drivers & install the correct Intel (or nVidia?) ones.

In a perfect world your FE h/w would match your MBE h/w when setting up a diskless booting system, however this is not necessary but when they don't match you will have to do manual fiddling like this to get the h/w running. Same may apply to your sound card drivers for example.

I also get the pacman nfs error & never tried to debug it as it seems to work fine (I'm assuming the 'already mounted' part of the error is what applies to me).

phisherpryce wrote:
The bootup process stops at 'Parent daemon running'. I'm not sure if the nfs errors are causing the problem or the video error.

Any ideas? Thanks.


No idea why it is hanging here. What's in /var/log/messages (accessible from your MBE under /nfsroot/<<FEName>>/var/log/messages, i.e., no need to boot up the FE to get this log).
Page 16 of 17 All times are UTC - 6 hours
Powered by phpBB® Forum Software © phpBB Group
http://www.phpbb.com/