Re: setup-storage: set vgname to hostname

[Jean Spirat]

I prefer using my own variables in my own namespace and initialize them
from a "known good" source - like $HOSTNAME. This may not be the most
obvious solution, but this way you can guarantee that "your" variable has
a valid value. And if $HOSTNAME turns out to be unrealiable - or undesired
- then you could easily change the source (like maybe retrieving the value
from an external database.) without having to wade through the rest of the
source code trying hard not to introduce new bugs.

tschüß
thomas
  
ok so any defined variables can be used in setup storage (like in 
http://www.informatik.uni-koeln.de/fai/fai-guide/ch-config.html#s-classvariables).


I had not thinked of this before i was seeing this as a static file. 
good to know thanks :)


regards,
JEan

Re: setup-storage: set vgname to hostname

[Jean Spirat]

Thomas Neumann a écrit :

You can write a hook that replaces $HOSTNAME with the current
hostname. This is the easiest way.



As a bonus: It even works!

;)

tschüß
thomas

  
do you know wich variable we could use in a way that is supported by FAI 
. I mean using a variable like $HOSTNAME or whatever would be great but 
only if this is a normal behavior and not an edge condition as if i 
upgrade FAI i do not want to have all my partition system stop working. 
Could you tell us your advice and variable list we could or should not use ?


regards,
JEan.

Re: Starting a FAI installation using kexec

[Jean Spirat]

I might not fully understand your case, but on an already
basic-installed system, I'd simply run fai softupdate.

Wasn't kexec some system for changing the running kernel without reboot?

Henning

  
i think the goal here is to take any machine from a provider that you do 
not control via dhcp or pxe and use fai anyway to reinstall it from scratch.


If you knwo how i am interested :)


regards,
Jean.

Re: FAI 3.2.20 released

[Jean Spirat]

Thomas Lange a écrit :

On Mon, 04 May 2009 13:26:55 +0200, Jean Spirat  said:



> is there any changelog somewhere ? Is gpt-bios included in this one ? :)
http://packages.debian.org/changelogs/pool/main/f/fai/current/changelog
but it's currently only for 3.2.19. Maybe tomorrow you will see the
changelog for 3.2.20 there.

  


yes i was there but .20 was not on it :)


In the meantime here's the changelog:

fai (3.2.20) unstable; urgency=low

  [ Thomas Lange ]
  * menu.lst: remove year
  * fai-cd: add option -B, that creates a boot-only CD (without containing
the nfsroot, the partitial mirror and the config space)
  * conf/menu.lst.boot-only: new config file for the boot only CD
  * fai-cd.8: add option -B
  * fai-abort: add missing lsb header
  * control: remove duplicate fields
  * copyright: fix lintian warning
  
  [ Michael Tautschnig ]

  * fixed several hyphen-used-as-minus-sign errors in man pages
  * update to standards version 3.8.1, no changes needed
  * replaced dh_clean -k by dh_prep
  * use dh_lintian to install lintian overrides
  * added override for fai-nfsroot (diverts foreign /etc/init.d.rcS)
  * Build-Depends: debhelper (>= 7)
  * setup-storage/Sizes.pm: Clarified error message in cases where the user
specified a partition is larger than the entire disk
  * setup-storage/Parser.pm, setup-storage/Sizes.pm, setup-storage/Volumes.pm:
Added pseudo-disklabel gpt-bios to support GPT on systems with BIOS
instead of EFI. Also installs an additional partition to store whatever
doesn't fit into MBR.
  * setup-storage: bumped version to 1.0.5 (gpt-bios support)

 -- Thomas Lange   Mon, 04 May 2009 10:42:25 +0200

  

great, thanks for this great work ! :)

regards,
Jean.

Re: FAI 3.2.20 released

[Jean Spirat]

Thomas Lange a écrit :

FAI 3.2.20 has been released. The command fai-cd has now a new option
-B, which creates a small boot-only CD, which does not contain the
nfsroot, the partitial mirror and the config space. It is meant for
booting from CD or USB stick but then the kernel gets all other parts
via network.

You will get the newest version if you add this line to the sources.list
file:

deb http://www.informatik.uni-koeln.de/fai/download lenny koeln
  

is there any changelog somewhere ? Is gpt-bios included in this one ? :)

Regards,
Jean.

Re: setup-storage 1.0.4 - ERROR (line 11): Invalid file: Wasexpecting /\Z/ but found

[Jean Spirat]

i agree that the error message could be a little different like: "disk  
is not big enough for the settings"
or "I don't want to talk to you no more, you empty headed animal food  
trough wiper. I fart in your general direction. Your mother was a  
hamster and your father smelt of elderberries", something that really  
told us we got somethign wrong in our configs ;)





I've changed the error message to

Sorry, can't create a partition of $start B on a disk of $size_b - check your 
config!

(with $start and $size_b being the size of the partition and that of the disk,
respectively). Will make it into the next experimental upload.

Best,
Michael

  
woohoo great, so much for the monthy python message then i guess this is 
easier to debug now ... ;)


regards,
Jean.

Re: setup-storage 1.0.4 - ERROR (line 11): Invalid file: Wasexpecting /\Z/ but found

[Jean Spirat]

You where soo right... Thnx... 

I was trying to assign 72304 x 1024 = 74039296 
disk_config disk1 bootable:1

primary /boot   256 ext3defaults
primary -   2048swaprw
primary -   7-  -   -

while I the disk as:
[r...@urpbe1 /root]# fdisk -l
Disk /dev/sda: 72.7 GB, 72746008576 bytes
255 heads, 63 sectors/track, 8844 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
 
Thnx a lot!


- vin
  
i agree that the error message could be a little different like: "disk 
is not big enough for the settings"
or "I don't want to talk to you no more, you empty headed animal food 
trough wiper. I fart in your general direction. Your mother was a 
hamster and your father smelt of elderberries", something that really 
told us we got somethign wrong in our configs ;)


regards,
jean

Re: setup-storage 1.0.4 - ERROR (line 11): Invalid file: Wasexpecting /\Z/ but found

[Jean Spirat]

Executing: mdadm --detail --scan --verbose -c partitions
Lower bound of partition size is greater than disk size
Error in task partition. Traceback: task_error task_partition task task_install 
task task_action task main

Usually n order to solve this I need to manually invoke a "dd if=/dev/zero 
of=/dev/sda bs=512 count=1" to remove any traces of a partition table then restart 
the install once more but this does not work on my Dell PE1950 ??

Any fix for this one perhaps ?

Help again greatly appreciated!

- vin

  
i had this issue once because i was allocating more storage than my disk 
has space for. Are you sure you are not overbooking the disk's capacity 
? If a dd let you go over this issue this should not be the case just 
giving experience return for the same error message i had..


try to put

vg00-root  /   5- xfs   defaults


and see if it helps...


regards,
Jean.

Re: setup-storage 1.0.4 - ERROR (line 11): Invalid file: Was expecting /\Z/ but found

[Jean Spirat]

hi,

i got the same issue, i think this is corrected in the experimental 
release. For me at least it was (with gpt-bios working as well) :)


regards,
Jean.

Re: gpt disk label and /boot

[Jean Spirat]

Ok; it'll take me a few days to come up with a better solution (I don't think we
can yet fully switch to grub-pc). Could you, in the meantime, just put some hack
into your scripts to set BOOT_DEVICE to /dev/md0 to see which further errors we
need to get fixed?

Thanks,
Michael

for what i tested it works if i force the boot_device to md0.

So there is left in my plate:

lvm options
md options

and then i will be out of idea to tourment you Michael ! :)

regards,
Jean.

Re: gpt disk label and /boot

[Jean Spirat]

Hmm, do you have any idea whether the version of grub that you are using has
proper support for RAID devices? setup-storage does some hackery in that it sets
BOOT_DEVICE to the actual devices that make up the RAID device because earlier
versions of grub had no idea about RAID and LVM. I think, newer versions should
support it properly, but that will then depend on the version of grub you are
acutally using, meaning that we can either modify setup-storage to set
BOOT_DEVICE to /dev/md0 or hack the script to call device2grub for each device
(sda, sdb).
  
i use grub_pc so it is grub2 ( 1.96) it support lvm/raid without any 
problem.



ii  grub-pc 
1.96+20080724-16 GRand Unified Bootloader, version 2 (PC/BIOS



regards,
Jean

Re: gpt disk label and /boot

[Jean Spirat]

That one should be fixed in 3.2.18+experimental4 which is just about to be built
and uploaded, but please be ready for more errors...

Thanks a lot for testing all the stuff,
Michael

  
this one worked for me with one issue only: in grub_pc the following 
line fails for mdx as it  says:



grub_dev=$(device2grub $BOOT_DEVICE)
if [ -z "$grub_dev" ] ; then
 grub_dev=$(echo $BOOT_DEVICE | sed 's#/dev/##g')
fi


if boot_device is /dev/md0 it give  (sda /dev/sdb) for grub_dev so it 
fails. I searched for device2grub but cannot find any clue about how it  
works.


in anycase the setup-storage part works for me.

regards,
Jean.

Re: gpt disk label and /boot

[Jean Spirat]

Sorry for the late reply; using disklabel:gpt-bios should finally work with
3.2.18+experimental3, at least the problem in the parser has been fixed :-)

Best,
Michael

  

i just tested it fails here:

sh-3.2# parted -s /dev/sdb mklabel gpt-bios

parted: invalid token: gpt-bios
Error: Expecting a disk label type.

regards,
Jean

Re: gpt disk label and /boot

[Jean Spirat]

hum in fact i am unable to make it work even with dos partitions...Is 
this an issue with this version ?


regards,
Jean.

Re: gpt disk label and /boot

[Jean Spirat]

Jean Spirat a écrit :



For now, I've only added the gpt-bios stuff. Could you please try
3.2.18+experimental2?

Thanks,
Michael

  

then i tried this:

disk_config /dev/sdadisklabel:gpt-bios
primary -   500 -   -
primary swap2G  swapsw
primary -   15G -   -
primary -   500--   -


idem for sdb and

disk_config raid
raid1   /boot   sda1,sdb1   ext3  -


and got:

Starting setup-storage 1.0.4

  ERROR (line 5): Invalid file: Was expecting /\Z/ but found 
"disk_config

  /dev/sda disklabel:gpt-bios " instead
Syntax error
Using config file: /var/lib/fai/config/disk_config/main-1


regards,
Jean.

Re: gpt disk label and /boot

[Jean Spirat]

For now, I've only added the gpt-bios stuff. Could you please try
3.2.18+experimental2?

Thanks,
Michael

  


ii  fai-client
3.2.18+experimental2  Fully Automatic 
Installation client package
ii  fai-doc   
3.2.18+experimental2  Documentation for FAI



i tried but i got the error:  Volume group gpt has not been declared yet.



i used this:

disk_config /dev/sdadisklabel:gpt
gpt-bios -  500 -   -
primary swap2G  swapsw
primary -   15G -   -
primary -   500--   -


regards,
Jean.

Re: netboot rescue disk with FAI

[Jean Spirat]

Most live cd's come with their own config and system for netbooting.
I don't see a benefit of doing this with FAI.
These systems most often have much more features than the FAI nfsroot
when you use it with sysinfo instead of install.


  


the goal here is to use the existing network infrastructure for this 
added functionality. parted, chkfs and such are allready not so bad for 
disaster recovery.




But how to you know _what_ to mount?
Assduming you have quite a bunch of different systems and disk
configurations, it's pretty hard to know for all of them what to mount.

In case of emergency, it should be enough to boot a system, and let the
person working on the disaster recovery/debugging decide what to mount.

Again, this is easier with for example the grml live cd, which at least
prepares some mountpoints for all disks/partitions it has found on the
system.

  


FAI with sysinfo seems to know it so i will try to use the system here 
and see if it can work for me.


thanks for the pointer i will see if i can use this in netboot for me if 
not FAI will be quite good by itself :)




One important feature would be to have a root password setup
that is different on each "netboot" etc... :)  Any ideas ?



Sorry, I don't understand anymore what you are trying to achieve here.

  


i want to be able to let each department to launch netboot rescue 
session of their servers and if they all have a standard password in 
netboot then anyone with ssh acces could connect to the live system so i 
want a new pass at each boot where my script know to who send the 
temporary netboot password. My hook should be able to do that i just 
have to be sure i do it before sshd start :)




In case of disaster, I'd want fast and easy access to a system, not
fiddle around with some magic random password.

Sure, that way anybody who can bring a system into this mode of
netbbot/rescue system, has full access to the system, but that is the
same with access to hardware...
  
the trick is that we are 230km away from the machines so i need to have 
everything done via ssh. As i want to delegate to department this i need 
to make sure only the department that is in charge of this particular 
server to be able to acces the live temporary system. So to be short i 
want to be sure that the one rescue booting the server is the one that 
can connect to it :)



regards,
JEan.

Re: netboot rescue disk with FAI

[Jean Spirat]

But don't go editing exactly this file :)

Write your own action, read the guide, and take this file as example.

Henning
  

Thanks !

yes this is what i want, take this as an exemple and create my hook with it.


regards,
Jean.

Re: netboot rescue disk with FAI

[Jean Spirat]

you can define actions on your own (for e.g. myaction) and boot them
with FAI_ACTION=myaction. Therefore create the file hooks/myaction
in the config space which will automatically be executed.

  

Great !

Can you help me find the script used by sysinfo so i can customize it to 
my needs ?


regards,
Jean.

netboot rescue disk with FAI

[Jean Spirat]

Hello,

I wanted to know if anyone use fAI to netboot a rescue disk on their 
system. I mean when a server is having issue and do not boot use the pxe 
boot and fai to use it to netboot a live system that will be used with 
parted and other filesystem tools to repair the system and chroot into 
/target etc ?


It seems sysinfo FAI_FLAGS is near what i look for, anyone know how to 
duplicate the sysinfo hook with for exemple mount RW instead of read 
only etc.. One important feature would be to have a root password setup 
that is different on each "netboot" etc... :)  Any ideas ?




regards,
Jean.

Re: gpt disk label and /boot

[Jean Spirat]

Most probably, this is currently _not_ the case (assuming you want the
BIOS-style version and not EFI). I guess a simple "parted print" should help you
verify this. If the partition is set to bios_grub, you are fine; if it has the
flag "boot", it's an EFI boot volume. I think this will best be fixed by adding
a specific disklabel "gpt-bios" to setup-storage that deals with such specifics.
Doing so is not really hard, I would just like to make sure that it will see
some testing. If you're willing to dive into this, I'd happily prepare a patch
in the next few days.

  



yes i can test it if this is in the next days no problem !

yes instead of primary for special case:

gpt-bios
gpt-gui:21686148-6449-6E6F-744E-656564454649

so we have a shortcut for the gpt-bios and can give any gid we want, dont know 
if this is usefull but...




Does that interfer in any way with the GPT stuff? Or is it just the possible
problems with grub?

[...]


no in fact this is stupidity from me the raid will not change anything 
to that... :)


regards,
Jean.

gpt disk label and /boot

[Jean Spirat]

Hi,

I have some issue with gpt and grub and i just wanted to know if the 
setup-storage configure the /boot with the gui 
21686148-6449-6E6F-744E-656564454649 (so that grub can use it). I do not 
know how to verify this at all some i am kind of lost :)


 of course things get complicated when the /boot  is on a raid/lvm 
partition. How do you handle this case ?


regards,
JEan.


The *BIOS Boot Partition* is a partition 
 on a data 
storage device  that 
may be used by standard BIOS -based 
machines in order to boot when the partition table of the device is a 
GPT  label.


The Globally Unique Identifier 
 for the BIOS 
Boot Partition in the GPT scheme is 
21686148-6449-6E6F-744E-656564454649. In the context of GPT labels, it 
is the BIOS equivalent of the EFI System Partition 
 (which is used by 
systems based on EFI  rather than BIOS).

lvm tags

[Jean Spirat]

hi,

is it possible to set lvm tags with setup storage ? From what i saw 
this is not but perhaps i overlooked something.


regards,
Jean.

setup-storage LVM / RAID options ideas

[Jean Spirat]

Hi,

 I was just wondering if this would be possible use  xxcreateopts="" on 
disk_config lvm and raid  to pass options to  it. For exemple we could 
have :



lvcreateopts="--stripes 4 --stripesize 256"
vgcreateopts="--physicalextentsize 2k"
pvcreateopts="--metadatacopies 1"


Same for raid to set the options like chunk size for raid 1 that i like:

mdcreateoption="--chunk=128"

for the ones who like to fine tune their file systems :)

 Do you think this is a good  idea for the future  ?

regards,
Jean.

-
exemple

disk_config sda

primary  -256 --  
primary  swap 1024swap   sw

primary  -  0-   --

disk_config sdb
primary  -  0-   - -

disk_config sdc
primary  -  0-   - -

disk_config sdd
primary  -  256 - -
primary  -  1024-   -
primary  -  0-   - -

disk_config sde
primary  -  0-   - -

disk_config sdf
primary  -  0-   - -

disk_config raid

raid1/sda1,sdd1  ext2rw,errors=remount-ro   
mdcreateoption="--chunk=128"
raid1swap sda2,sdd2  swaprw 
raid1-sda3,sdd3  ext2default


raid0-sdb1,sde1  ext2default
raid0-sdc1,sdf1  ext2default

# config the LVM
disk_config lvm pvcreateopts="--metadatacopies 1"
vg  my_pv   md2,md3   vgcreateopts="--physicalextentsize 2k"
my_pv-_usr	/usr			2048ext3  rw -O dir_index,resize_inode 
lvcreateopts="--stripes 4 --stripesize 256" 
my_pv-_var	/var			600 ext3  rw -O dir_index,resize_inode

my_pv-_e_h  /export/home10240   reiser  rw,notail
my_pv-_e_s  /export/sites   2048reiser  rw,notail
my_pv-_v	/vservers		2048ext3rw 

Re: using varaibles in setup-storage

[Jean Spirat]

Jean Spirat a écrit :

hi,

I just wanted to know if we can use variables in setup-storage config 
files ? I wanted to use the hostname to name my vg group  and i 
wondered if this was possible.


regards,
Jean
it seems this do not work at least with $HOTNAME for exemple, perhaps 
any other way (env variables or other ?)  ?


regards,
Jean.

using varaibles in setup-storage

[Jean Spirat]

hi,

I just wanted to know if we can use variables in setup-storage config 
files ? I wanted to use the hostname to name my vg group  and i wondered 
if this was possible.


regards,
Jean

Re: setup-storage and AOE => fails at Use of uninitialized value in multiplication (*) at /usr/share/fai/setup-storage//Sizes.pm line 618.

[Jean Spirat]

Ok, I've added the suggested patches (with minor changes) to the experimental
tree; could you please test 3.2.17+experimental2 (see
http://faiwiki.debian.net/index.php/Main_Page#getting_FAI for info how to obtain
the experimental builds).

Thanks a lot,
Michael

  


r...@aqded01:/# dpkg -l|grep fai
ii  fai-client  3.2.17+experimental3  Fully 
Automatic Installation client package
ii  fai-nfsroot 3.2.17+experimental3  Fully 
Automatic Installation nfsroot package




   ERROR: Internal error in generated parser code!
   (Hint: Global symbol "$vol_id" requires explicit package 
name

  at (eval 52) line 4279. BEGIN not safe after errors--
  compilation aborted at (eval 52) line 4425. )
setup-harddisks is running in test-only mode
Starting setup-storage 1.0.4
Can't call method "file" on an undefined value at 
/usr/share/fai/setup-storage//Parser.pm line 714, <$config_file> line 1.

disklist was: etherd/e10.0
etherd/e10.1
etherd/e10.2
sda



i sent you directly the debug output i got.

Jean.

Re: setup-storage and AOE

[Jean Spirat]

hello michael,

just wanted to ping you just to know if you think the little chnage i 
made for AOE seems ok to you and have any chance to be in a future 
release. I know you are quite busy it is just to know if you saw the 
thread :)


Thanks again for your work it install beautifully now.

regards,
Jean.

Re: GPT disklabels

[Jean Spirat]

my bad it seems this was an typo in the disk config that bring up this 
error. Gpt is working fine with udev :)


Mounting /dev/sda4 to /target/boot
[  349.653872] EXT3-fs: sda6: couldn't mount because of unsupported 
optional features (40).

mount: wrong fs type, bad option, bad superblock on /dev/sda4,
  missing codepage or helper program, or other error
  In some cases useful info is found in syslog - try
  dme[  349.680921] EXT3-fs: sda7: couldn't mount because of 
unsupported optional features (40).

sg | tail  or so


the error is that the system was not able to mkfs and mount the 
different disk.


regards,
Jean

GPT disklabels

[Jean Spirat]

hi,

 It seems that setup storage support gpt partition type but from my 
test linux do not like it as udev utils do not see the disk correctly 
leading to fai problems:




Parted works:

Executing: udevsettle --timeout 10 && parted -s /dev/sda mklabel gpt
Executing: udevsettle --timeout 10 && parted -s /dev/sda mklabel gpt
Executing: udevsettle --timeout 10 && parted -s /dev/sda mkpart primary 
linux-swap 17408B 4294984703B
Executing: udevsettle --timeout 10 && parted -s /dev/sda mkpart primary 
ext3 4294984704B 15032402943B
Executing: udevsettle --timeout 10 && parted -s /dev/sda mkpart primary 
ext3 15032402944B 15346975743B
Executing: udevsettle --timeout 10 && parted -s /dev/sda mkpart primary 
ext3 15346975744B 26084393983B
Executing: udevsettle --timeout 10 && parted -s /dev/sda mkpart primary 
ext3 26084393984B 36821812223B
Executing: udevsettle --timeout 10 && parted -s /dev/sda mkpart primary 
ext3 36821812224B 42190521343B
Executing: udevsettle --timeout 10 && parted -s /dev/sda mkpart primary 
ext3 42190521344B 24982591B

Executing: udevsettle --timeout 10 && parted -s /dev/sda set 1 boot on

but  /lib/udev/vol_id  fail and fai stop on this error:


Executing: /lib/udev/vol_id -u /dev/sda6
Command /lib/udev/vol_id -u /dev/sda6 had exit code 4
Failed to obtain UUID for /dev/sda6
Traceback: task_error task_partition task task_install task task_action 
task main

FATAL ERROR. Installation stopped.


when you test:


r...@aqded01:/# /lib/udev/vol_id -u /dev/sda6
/dev/sda6: unknown volume type
r...@aqded01:/# /lib/udev/vol_id -u /dev/sda5
/dev/sda5: unknown volume type
r...@aqded01:/# /lib/udev/vol_id -u /dev/sda4
/dev/sda4: unknown volume type
r...@aqded01:/# /lib/udev/vol_id -u /dev/sda3
/dev/sda3: unknown volume type
r...@aqded01:/# /lib/udev/vol_id -u /dev/sda1
6285ffb7-b04f-4287-8a33-f2b8a80861c3

so basicaly udev utils seems limited to msdos partition type , Anyone 
ever used GPT style partitions they seems more robust the msdos ones  ?




regards,
Jean.

Re: setup-storage and AOE => fails at Use of uninitialized value in multiplication (*) at /usr/share/fai/setup-storage//Sizes.pm line 618.

[Jean Spirat]

to isolate the error message from the complete log this is:

INTERNAL ERROR in setup-storage:
end < start
Please report this error to the Debian Bug Tracking System.

it was just that my test partition was too small for the config i had, 
the message is not very clear :)



in /usr/share/fai/setup-storage/Sizes.pm i suggest a change from  ligne 
77/706:

($end >= $start) or &FAI::internal_error("end < start");
to
($end >= $start) or &FAI::internal_error("end < start  $end < $start, is 
the sum of the partitions size not > to the size of your disk ?");






So to wrapup, here are my changes for AOE support in setup-storage:

/usr/share/fai/setup-storage/Init.pm



#
# @brief Check, whether $dev is a physical device, and extract sub-parts
#
# @param $dev Device string
#
# @return 1, if it the matches the regexp, and disk device string, and
# partition number, if any, otherwise -1
#

sub phys_dev {
 my ($dev) = @_;
 ( $FAI::debug > 0 ) and print "--- debug : $dev\n";
 if ($dev =~ m{^/dev/(etherd/e\d+\.\d+)p(\d+)?$})
 {
   ( $FAI::debug > 0 ) and print "--- debug : $dev - $1 - $2 .\n";
   defined("$2") or return (1, "/dev/$1", -1);
   return (1, "/dev/$1", $2);
 }
 elsif ($dev =~ m{^/dev/(i2o/hd[a-t]|sd[a-t]{1,2}|hd[a-t]|vd[a-t])(\d+)?$})
 {
   defined($2) or return (1, "/dev/$1", -1);
   return (1, "/dev/$1", $2);
 }
 elsif ($dev =~ 
m{^/dev/(cciss/c\dd\d|ida/c\dd\d|rd/c\dd\d|ataraid/d\d)p(\d+)?$})

 {
   defined($2) or return (1, "/dev/$1", -1);
   return (1, "/dev/$1", $2);
 }
 return (0, "", -2);
}



later in the same file:



#
# @brief Convert a device name and a partition id to a proper device name,
# handling cciss and the like
#
# @param $dev Device name of disk
# @param $id Partition id
#
# @return Full device name
#

sub make_device_name {

 my ($dev, $p) = @_;


 $dev .= "p" if ($dev =~
   m{^/dev/(cciss/c\dd\d|ida/c\dd\d|rd/c\dd\d|ataraid/d\d)$});

 $dev .= "p" if ($dev =~
   m{^/dev/(etherd/e\d+\.\d+)$} );

 ( $FAI::debug > 0 ) and print "--- debug dev,part: $dev - $p\n";

 $dev .= $p;
 internal_error("Invalid device $dev") unless (&FAI::phys_dev($dev))[0];
 return $dev;
}



and also in /usr/lib/fai/disk-info:


changed the regex to:

egrep ' etherd/e[[:digit:]]+\.[[:digit:]]+\b| i2o/hd.\b| cciss/c.d.\b| 
ida/c.d.\b| rd/c.d.\b| hd.\b| sd[a-z]{1,2}\b|/disc\b| vd.\b' 
/proc/partitions | diskandsize




hope this will help the community. Thanks for the FAI product it rocks ! :)

Jean.

Re: setup-storage and AOE => fails at Use of uninitialized value in multiplication (*) at /usr/share/fai/setup-storage//Sizes.pm line 618.

[Jean Spirat]

ok so now i got the etherdrive disk listed in the debug output but all 
fail nonetheless so i am lost at why the tool fail, here are the logs if 
it helps.


regards,
Jean.





Calling task_install
Calling task_partition
Partitioning local harddisks using setup-storage
Starting setup-storage 1.0.3
disklist was: etherd/e10.0
etherd/e10.1
etherd/e10.2
sda
Using config file: /var/lib/fai/config/disk_config/FAIBASE
Input was:
# example of new config file for setup-storage
#
#

disk_config /dev/etherd/e10.1 disklabel:msdos

primary swap16G swaprw
primary /   10G ext3rw,noatime
primary /temporary  50G-ext3rw,noatime

(CMD) parted -s /dev/etherd/e10.0 unit TiB print 1> /tmp/wyI8UzLwfK 2> 
/tmp/4qy5rod4so

Executing: parted -s /dev/etherd/e10.0 unit TiB print
(STDOUT) Model: Unknown (unknown)
(STDOUT) Disk /dev/etherd/e10.0: 0.27TiB
(STDOUT) Sector size (logical/physical): 512B/512B
(STDOUT) Partition Table: msdos
(STDOUT)
(STDOUT) Number  StartEnd  Size Type File system  Flags
(STDOUT)  1  0.00TiB  0.27TiB  0.27TiB  primary  ext3
(STDOUT)
(CMD) parted -s /dev/etherd/e10.0 unit B print free 1> /tmp/CjkgYSBcQy 
2> /tmp/5S51ldLUi2

Executing: parted -s /dev/etherd/e10.0 unit B print free
(STDOUT) Model: Unknown (unknown)
(STDOUT) Disk /dev/etherd/e10.0: 31787904B
(STDOUT) Sector size (logical/physical): 512B/512B
(STDOUT) Partition Table: msdos
(STDOUT)
(STDOUT) Number  Start   EndSize   Type File 
system  Flags

(STDOUT)  1  32256B  30637439B  30605184B  primary  ext3
(STDOUT)
(CMD) parted -s /dev/etherd/e10.0 unit chs print free 1> /tmp/1QU9Gdf9pM 
2> /tmp/rSWjqSkDTK

Executing: parted -s /dev/etherd/e10.0 unit chs print free
(STDOUT) Model: Unknown (unknown)
(STDOUT) Disk /dev/etherd/e10.0: 36473,35,41
(STDOUT) Sector size (logical/physical): 512B/512B
(STDOUT) BIOS cylinder,head,sector geometry: 36473,255,63.  Each 
cylinder is 8225kB.

(STDOUT) Partition Table: msdos
(STDOUT)
(STDOUT) Number  Start  End   Type File system  Flags
(STDOUT)  1  0,1,0  36472,254,62  primary  ext3
(STDOUT)
(CMD) parted -s /dev/etherd/e10.1 unit TiB print 1> /tmp/wrSEvuMyj7 2> 
/tmp/mtMwaTzxPL

Executing: parted -s /dev/etherd/e10.1 unit TiB print
(STDOUT) Model: Unknown (unknown)
(STDOUT) Disk /dev/etherd/e10.1: 0.05TiB
(STDOUT) Sector size (logical/physical): 512B/512B
(STDOUT) Partition Table: msdos
(STDOUT)
(STDOUT) Number  StartEnd  Size Type File system  Flags
(STDOUT)  1  0.00TiB  0.03TiB  0.03TiB  primary  ext3
(STDOUT)  2  0.03TiB  0.03TiB  0.00TiB  primary  ext3
(STDOUT)
(CMD) parted -s /dev/etherd/e10.1 unit B print free 1> /tmp/LUvw7BIfqL 
2> /tmp/y6j1MnNZJj

Executing: parted -s /dev/etherd/e10.1 unit B print free
(STDOUT) Model: Unknown (unknown)
(STDOUT) Disk /dev/etherd/e10.1: 5297984B
(STDOUT) Sector size (logical/physical): 512B/512B
(STDOUT) Partition Table: msdos
(STDOUT)
(STDOUT) Number  Start End   Size  Type File 
system  Flags

(STDOUT)  1  32256B30005821439B  30005789184B  primary  ext3
(STDOUT)  2  30005821440B  35015016959B  5009195520B   primary  ext3
(STDOUT) 35015016960B  49993251839B  14978234880B   Free 
Space

(STDOUT)
(CMD) parted -s /dev/etherd/e10.1 unit chs print free 1> /tmp/S8x3NXtjZq 
2> /tmp/IgGIvWm0BD

Executing: parted -s /dev/etherd/e10.1 unit chs print free
(STDOUT) Model: Unknown (unknown)
(STDOUT) Disk /dev/etherd/e10.1: 6078,218,27
(STDOUT) Sector size (logical/physical): 512B/512B
(STDOUT) BIOS cylinder,head,sector geometry: 6078,255,63.  Each cylinder 
is 8225kB.

(STDOUT) Partition Table: msdos
(STDOUT)
(STDOUT) Number  Start End  Type File system  Flags
(STDOUT)  1  0,1,0 3647,254,62  primary  ext3
(STDOUT)  2  3648,0,0  4256,254,62  primary  ext3
(STDOUT) 4257,0,0  6077,254,62   Free Space
(STDOUT)
(CMD) parted -s /dev/etherd/e10.2 unit TiB print 1> /tmp/73a3HOfigK 2> 
/tmp/ES8eUdhBj4

Executing: parted -s /dev/etherd/e10.2 unit TiB print
(STDOUT) Model: Unknown (unknown)
(STDOUT) Disk /dev/etherd/e10.2: 0.00TiB
(STDOUT) Sector size (logical/physical): 512B/512B
(STDOUT) Partition Table: msdos
(STDOUT)
(STDOUT) Number  Start  End  Size  Type  File system  Flags
(STDOUT)
(CMD) parted -s /dev/etherd/e10.2 unit B print free 1> /tmp/cLV41wlN3H 
2> /tmp/llEMb9q4xG

Executing: parted -s /dev/etherd/e10.2 unit B print free
(STDOUT) Model: Unknown (unknown)
(STDOUT) Disk /dev/etherd/e10.2: 1002438656B
(STDOUT) Sector size (logical/physical): 512B/512B
(STDOUT) Partition Table: msdos
(STDOUT)
(STDOUT) Number  Start  End  Size Type  File system  Flags
(STDOUT) 0B 1002438655B  1002438656BFree Space
(STDOUT)
(CMD) parted -s /dev/etherd/e10.2 unit chs print free 1> /tmp/RVSx43p5zq 
2> /tmp/lupHArIbmV

Executing: parted -s /dev/etherd/e10.2 unit chs print free
(STDOUT) Mod

Re: setup-storage and AOE => fails at Use of uninitialized value in multiplication (*) at /usr/share/fai/setup-storage//Sizes.pm line 618.

[Jean Spirat]

Jean Spirat a écrit :

i also found in

/usr/lib/fai/disk-info

a regex where i added etherd:

egrep ' etherd/e| i2o/hd.\b| cciss/c.d.\b| ida/c.d.\b| rd/c.d.\b| 
hd.\b| sd[a-z]{1,2}\b|/disc\b| vd.\b' /proc/partitions | diskandsize



is that enough ?

regards,
Jean.

i changed to
egrep ' etherd/e[[:digit:]]+\.[[:digit:]]+\b| i2o/hd.\b| cciss/c.d.\b| 
ida/c.d.\b| rd/c.d.\b| hd.\b| sd[a-z]{1,2}\b|/disc\b| vd.\b' 
/proc/partitions | diskandsize


i continue to search, any help would be appreciated ;)

regards,
Jean.

Re: setup-storage and AOE => fails at Use of uninitialized value in multiplication (*) at /usr/share/fai/setup-storage//Sizes.pm line 618.

[Jean Spirat]

i also found in

/usr/lib/fai/disk-info

a regex where i added etherd:

egrep ' etherd/e| i2o/hd.\b| cciss/c.d.\b| ida/c.d.\b| rd/c.d.\b| hd.\b| 
sd[a-z]{1,2}\b|/disc\b| vd.\b' /proc/partitions | diskandsize



is that enough ?

regards,
Jean.

Re: setup-storage and AOE => fails at Use of uninitialized value in multiplication (*) at /usr/share/fai/setup-storage//Sizes.pm line 618.

[Jean Spirat]

I tried to change the file:

/usr/share/fai/setup-storage/Init.pm

i modified the regex in this part:  :

sub phys_dev {
 my ($dev) = @_;
 if ($dev =~ m{^/dev/(etherd/e\d+\.\d+)p(\d+)?$})
 {
   defined("$2") or return (1, "/dev/$1", -1);
   return (1, "/dev/$1", $2);
 }
 elsif ($dev =~ m{^/dev/(i2o/hd[a-t]|sd[a-t]{1,2}|hd[a-t]|vd[a-t])(\d+)?$})
 {
   defined($2) or return (1, "/dev/$1", -1);
   return (1, "/dev/$1", $2);
 }
 elsif ($dev =~ 
m{^/dev/(cciss/c\dd\d|ida/c\dd\d|rd/c\dd\d|ataraid/d\d)p(\d+)?$})

 {
   defined($2) or return (1, "/dev/$1", -1);
   return (1, "/dev/$1", $2);
 }
 return (0, "", -2);
}


so etherd device can be seen but it seems to not be enough (i am far 
from understanding the code i just try to make the regex work if it find 
a etherd device :)  ).



running parted on the AOE target give same results as on a /dev/sda 
target so i do not see where is the issue :(


regards,
Jean

setup-storage and AOE => fails at Use of uninitialized value in multiplication (*) at /usr/share/fai/setup-storage//Sizes.pm line 618.

[Jean Spirat]

Hello,

 To continue the AOE experience now FAI works and start but it fail if 
i put an AOE storage disk:


Calling task_install
Calling task_partition
Partitioning local harddisks using setup-storage
Starting setup-storage 1.0.3
Using config file: /var/lib/fai/config/disk_config/FAIBASE
Executing: parted -s /dev/sda unit TiB print
Executing: parted -s /dev/sda unit B print free
Executing: parted -s /dev/sda unit chs print free
Executing: vgdisplay --units m -s
Executing: mdadm --detail --scan --verbose -c partitions
Use of uninitialized value in multiplication (*) at 
/usr/share/fai/setup-storage//Sizes.pm line 618.
Use of uninitialized value in multiplication (*) at 
/usr/share/fai/setup-storage//Sizes.pm line 618.
Use of uninitialized value in multiplication (*) at 
/usr/share/fai/setup-storage//Sizes.pm line 622.
Use of uninitialized value in multiplication (*) at 
/usr/share/fai/setup-storage//Sizes.pm line 622.
Use of uninitialized value in concatenation (.) or string at 
/usr/share/fai/setup-storage//Sizes.pm line 425.

INTERNAL ERROR in setup-storage:
convert_unit B
Please report this error to the Debian Bug Tracking System.
Traceback: task_error task_partition task task_install task task_action 
task main

FATAL ERROR. Installation stopped.
bash: no job control in this shell


here is the config file i use:

disk_config /dev/etherd/e10.1 disklabel:msdos

primary swap16G swaprw
primary /   10G ext3rw,noatime
primary /temporary  50G-ext3rw,noatime


if i put /dev/sda it works but with the aoe target  it fails. It seems 
to fail here:


 $next_start = $current_disk->{bios_sectors_per_track} *
   $current_disk->{sector_size};

Any idea ?


regards,
Jean

Re: config space using GIT

[Jean Spirat]

ok error found, do NOT call you directory 'var/www/nginx-default/config/ 
', i changed to 'configuration' and now it works...


Jean.

config space using GIT

[Jean Spirat]

Hello,

i tried to use GIT for the config file path. I used:

FAI_CONFIG_SRC=git+http://10.1.0.251/config


and in the documentroot is :

fai01:/# ll /var/www/nginx-default/config/
total 32
drwxr-xr-x  2 root root 4096 jan 28 00:31 branches
-rw-r--r--  1 root root   66 jan 28 00:31 config
-rw-r--r--  1 root root   58 jan 28 00:31 description
-rw-r--r--  1 root root0 jan 28 00:32 git-daemon-export-ok
-rw-r--r--  1 root root   23 jan 28 00:31 HEAD
drwxr-xr-x  2 root root 4096 jan 28 00:31 hooks
drwxr-xr-x  2 root root 4096 jan 28 00:32 info
drwxr-xr-x 64 root root 4096 jan 28 00:31 objects
drwxr-xr-x  4 root root 4096 jan 28 00:31 refs



  But at boot i allways have this error:

Calling task_confdir
Kernel currently running: Linux 2.6.26-1-amd64 x86_64 GNU/Linux
Kernel parameters: initrd=initrd.img-2.6.26-1-amd64 ip=dhcp boot=live  
FAI_FLAGS=verbose,createvt FAI_ACTION=install  debug BOOT_IMAGE=vmlinuz-

Reading /tmp/fai/boot.log
FAI_FLAGS: verbose createvt
git+http
Checking out from git
fatal: destination directory '/var/lib/fai/config' already exists.
WARNING: directory /var/lib/fai/config/class not found.
Problems accessing the config space.


I took care of not having /var/lib/fai/config in the nfsroot so i do 
not know why i have this error. Perhaps a little missconfiguration from 
me :)


Regards,
Jean.

Re: FAI and debian live

[Jean Spirat]

ok wrong list sorry :)

Jean.

Re: FAI and debian live

[Jean Spirat]

a forward from some talk with the FAI guys.. perhaps you can help me here :)




FAI only needs the live-initramfs package which provides the functions
to boot from network, mounts the root file system via NFS and adds 
aufs (another unionfs) on top of it. Using aufs we have write access to all

files inside the nfsroot which was mounted read-only.
Other feature of live-initramfs (like setup the locale, enable a user
account,...) are not needed by FAI.

live-initramfs also is usefull for the fai-cd, were it also helps
during the boot process.

There are no plans to use more features of debian live. I heared that
live-initramfs will be split into more packages in the future, so we
may need less parts from it.

  

ok thanks :)

in fact this is where my AOE mount fails, the system ends up read only
and the aufs part is not working.

What should i take care about when i mount the root file system so it fits with 
the live system ?

For now i just pur ROOT=/dev/etherd/xxx in the initramfs.conf and use the initrd
to bring the network and AOE tools up (in init-premount).  From the
prompt of the FAI boot i can do a mount and see that there is a 'rootfs'
for / and then my AOE mount also for / so i bet there is only a little
thing i need to trwich to get it done no ?


regards,
Jean.

FAI and debian live

[Jean Spirat]

Hi,

 I wanted to know how is FAI using debian live. It seems to use the 
debian live-initramfs system but i wanted to know if you use anything 
else from the debian live project and how you interact with it :) Any 
little information you could provide ?


regards,
Jean.

Re: NFSROOT, for AOE booting

[Jean Spirat]

are you sure, that the initrd you are booting via pxe is really the
one you have changed?
I also need changes to the initrd and as there is currently no established
method in FAI to do this. I use NFSROOT_HOOKS to run a
script at the end of make-fai-nfsroot, which sets up the initramfs hooks,
re-runs update-initramfs and copies the results again. Not the most elegant
way, but I don't see a better one at the moment.

This is the hook file I use:
# cat /etc/fai/nfsroot-hooks/patch-initrd

#!/bin/sh
#

echo "Enter patch initrd hook"

# copy files to patch initrd
cp /usr/local/etc/*patch $NFSROOT/usr/local/etc
cp /usr/local/etc/initramfs-tools_hook_network 
$NFSROOT/etc/initramfs-tools/hooks/network

$ROOTCMD update-initramfs -k all -t -u
cp -pv $NFSROOT/boot/vmlinu?-* $NFSROOT/boot/initrd.img-* $TFTPROOT

echo "Exit patch initrd hook"


Bye, Ralf
  
thanks, i will try this one. The initrd is the good one, For now i am 
able to boot from aoe but the root fs is mounted read only :) so FAI 
fails to start.





FAI_ACTION: install
Performing FAI installation. All data may be overwritten!

Calling task_install
/usr/lib/fai/subroutines: line 470: 
/var/run/fai/FAI_INSTALLATION_IN_PROGRESS: Read-only file system
/usr/lib/fai/subroutines: line 86: /tmp/fai/dmesg.log: No such file or 
directory

Calling task_partition
Partitioning local harddisks using setup_harddisks
tee: /tmp/fai/format.log: No such file or directory
setup_harddisks version 0.43fai
Probing disks:
Disks found: sda etherd/e10.0 etherd/e10.1

ERROR: no config file for setup_harddisk found. Please check you classes 
and files in disk_config.

cat: /tmp/fai/format.log: No such file or directory
mkdir: cannot create directory `/target': Read-only file system
mkdir: cannot create directory `/target': Read-only file system
find: `/tmp/fai': No such file or directory
ln: creating symbolic link `/tmp/fai/../last-install': No such file or 
directory

ln: creating symbolic link `/tmp/fai/../last': No such file or directory
mkdir: cannot create directory `/target': Read-only file system
cp: cannot stat `/tmp/fai/*': No such file or directory
chown: cannot access `/target/var/log/fai/(none)/install-': No such file 
or directory
chgrp: cannot access `/target/var/log/fai/(none)/install-': No such file 
or directory
chmod: cannot access `/target/var/log/fai/(none)/install-': No such file 
or directory
ln: creating symbolic link `/target/var/log/fai/localhost': No such file 
or directory
ln: creating symbolic link `/target/var/log/fai/(none)/last-install': No 
such file or directory
ln: creating symbolic link `/target/var/log/fai/(none)/last': No such 
file or directory

Partitioning tool did not create /tmp/fai/disk_var.sh file.
bash: no job control in this shell
root@(none):/# mount
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,size=10240k,mode=755)
/dev/etherd/e10.1p1 on / type ext3 (ro,errors=continue,data=ordered)


i have to digg a little to find why it is mounted read only and then the 
system will be able to work.


regards,
Jean

Re: NFSROOT, for AOE booting

[Jean Spirat]

I guess it would be interesting for FAI if we could claim to work with
AOE SAN's - but I also guess you will not get too much help with it -
mainly lacking experience, not willingness to help - I didn't see AOE
being mentioned here before.
  


yes thanks for your insight on this.

I agree that this is not a setup that is seen everywhere and i would 
really like to see FAI support this i am sure this is not so much 
complicated but i lack knowledge on the live initramsfs part.
Do you know where is the best place to get knowledgeable people for this 
particular part ?


I think if aoe is working we could quickly also add iscsi and other san 
system.


regards,
Jean.

Re: NFSROOT, for AOE booting

[Jean Spirat]

Thomas Lange a écrit :

On Tue, 16 Dec 2008 18:03:44 +0100, Jean Spirat  said:



>   I am trying to build a nfsroot that will be mounted using AOE and not 
> NFS. For this i use a partition mounted here 
> /srv/fai/nfsroot/live/filesystem.dir. The issue i have is that if i 
> rebuild the nfsroot i got an error: /usr/sbin/make-fai-nfsroot tries to 
> remove the mountpoint and fail. Is there any way to go around that ?

Please give us the exact error message of make-fai-nfsroot -v

Why not NFS? Try to keep things simple. NFS will work without any
problems. 
  

fai01:/# make-fai-nfsroot
Creating FAI nfsroot in /srv/fai/nfsroot/live/filesystem.dir.
By default it needs more than 390 MBytes disk space.
This may take a long time.
/srv/fai/nfsroot/live/filesystem.dir already exists. Removing 
/srv/fai/nfsroot/live/filesystem.dir
rm: ne peut détruire le répertoire 
`/srv/fai/nfsroot/live/filesystem.dir': Périphérique ou ressource occupé

Log file written to /var/log/fai/make-fai-nfsroot.log

Basicaly it want to remove the mountpoint.

NFS is considered unsecure by some, it is complicated, it open plenty of 
ports on your host and a single config error would make my life a 
security hell (installing corrupted server images), i am in an open 
infrastructure so the less i open it, the better i feel,


Also for linux the nfs daemon is located for a big part in kernel so it 
cannot be run in vserver guests ( user space version is completly 
obsolete) and this is an issue for me nothing should run on the host, 
andi have an AOE target so why not use it as  it remove all those 
troublesome issues for me in one swipe.



The issue is then how to create all the initramfs files and make the 
make-fai-nfsroot works with:


- custom initramfs configuration
- nfsroot mount point

without needing manual tweaking for it :) 



Also i have issues because initramfs cannot mount the /live I have

/srv/fai/nfsroot/live/filesystem.dir/etc/initramfs-tools/scripts/local-top/aoetools
/srv/fai/nfsroot/live/filesystem.dir/etc/initramfs-tools/hooks/aoetools

the hooks copy the module and aoe-discover tool, the localtop should 
"ifconfig up" all the interfaces and aoe-discover to create the 
/dev/ether/ex device but it seems the /live is done before. Any one got 
links to some help or know how to solve this ?


i am stuck at the initramsfs prompt and i can modprobe aoe and 
aoe-discover and it works :




BusyBox v1.10.2 (Debian 1:1.10.2-2) built-in shell (ash)
Enter 'help' for a list of built-in commands.

/bin/sh: can't access tty; job control turned off
(initramfs)
(initramfs) ls -la /dev/tty
crw-rw-rw-1 00  5,   0 Dec 17 14:36 /dev/tty
(initramfs) modprobe aoe
(initramfs) ifconfig eth3 up
(initramfs) aoe-discover
(initramfs) [ 2804.565557] aoe: 0030486265aa e10.0 v00a6 has 585940992 
sectors

[ 2804.565557] aoe: 0030486265aa e10.1 v00a6 has 97656832 sectors
[ 2804.565557]  etherd/e10.0: p1
[ 2804.598817]  etherd/e10.1: p1 p2


but it does not do it automaticaly to mount and boot the FAI system.

regards,
Jean.

NFSROOT, for AOE booting

[Jean Spirat]

Hi,

 I am trying to build a nfsroot that will be mounted using AOE and not 
NFS. For this i use a partition mounted here 
/srv/fai/nfsroot/live/filesystem.dir. The issue i have is that if i 
rebuild the nfsroot i got an error: /usr/sbin/make-fai-nfsroot tries to 
remove the mountpoint and fail. Is there any way to go around that ?


regards,
Jean.

making custom mkinitramfs

[Jean Spirat]

Hi,

I wanted to know the best way to make a custom mkinitramfs for FAI. 
Right now i just changed my configs and used:


mkinitramfs -o /srv/tftp/fai/initrd.img-2.6.26-1-amd64  2.6.26-1-amd64


 Is it enough ? I guess not as the unionfs and such should be installed 
somewhere no ?


regards,
JEan.

Re: NFS/dhcp/tftp on eth1, kernel search NFS it on eth0

[Jean Spirat]

Jean Spirat a écrit :

Hello,

 All my tests works well but i have a failure when i try to install a 
host with two ethernet cards. The dhcp and tftp runs on the eth1 
network and it works fine. But the issue is that the fai kernel when 
loaded bring not eth1 but eth0.


 So it can never connect to the nfs server.i tried to add 
ether=0,0,eth1 to the pxe config but this does not work as it seems 
the tg3 modules is not in kernel but comes from the initramfs. I do 
not see how i can change anything in the fai kernel or initd to make 
this works. Anyone know how to deal with this problem or know if i do 
something wrong ?


regards,
Jean.

to add a little more explanation:


- host boot ok
- pxe boot get the ip from the dhcp
- tftp load the kernel ok
- kernel fail to get the NFS server as it brings only eth0 up

Regards,
Jean.

NFS/dhcp/tftp on eth1, kernel search NFS it on eth0

[Jean Spirat]

Hello,

 All my tests works well but i have a failure when i try to install a 
host with two ethernet cards. The dhcp and tftp runs on the eth1 network 
and it works fine. But the issue is that the fai kernel when loaded 
bring not eth1 but eth0.


 So it can never connect to the nfs server.i tried to add 
ether=0,0,eth1 to the pxe config but this does not work as it seems the 
tg3 modules is not in kernel but comes from the initramfs. I do not see 
how i can change anything in the fai kernel or initd to make this works. 
Anyone know how to deal with this problem or know if i do something wrong ?


regards,
Jean.

Re: it work, ..allmost ! :) SOLVED

[Jean Spirat]

Michael Tautschnig a écrit :

[...]
  

anything special about the initramsfs that i should do ?




Did you try adding

$ROOTCMD update-initramfs -k all -u

to scripts/LAST/60-misc.sh? Don't know whether this helps or not, but it might
be worth a try.

Best,
Michael

  

it works now:

things learned:

- Add the raid and lvm package to the installed server's packages
- Do not forget to chmod 755 the script you create in 
/src/fai/config/scripts !

- Use grub-pc and not grub legacy on lenny

Thanks a lot especialy to Ryan for removing the last standing stone in 
the path !



here is my script:

/srv/fai/config/scripts/LAST/60-misc


   #!/bin/bash

   $ROOTCMD /usr/bin/grub-mkimage --output=/boot/grub/core.img ext2 pc
   gpt biosdisk lvm
   $ROOTCMD /usr/sbin/grub-install '(md0)'
   $ROOTCMD /usr/sbin/grub-setup   '(md0)'
   $ROOTCMD /usr/sbin/update-grub
   $ROOTCMD update-initramfs -k all -u



i also changed all grub package name to grub-pc and i removed the files 
related to grub from the '/files/' directory of the exemple ones.



Jean.

Re: it work, ..allmost ! :)

[Jean Spirat]

As Thomas already said: Please post (at least) the error message as reported by
setup-storage -- I don't see any obvious error in the config you posted.

Thanks,
Michael

  
i think all the issue comes from the grub install. I switched to grub2 
as it better understand the raid settings and lvm settings.



I tried this:

echo 'GRUB install, mkimage';
chroot /target /usr/bin/grub-mkimage --output=/boot/grub/core.img ext2 
pc gpt biosdisk lvm ;

ls  /target/boot/grub;

echo 'grub-install';
chroot /target /usr/sbin/grub-install '(md0)';
ls  /target/boot/grub;

echo 'grub-setup';
chroot /target /usr/sbin/grub-setup   '(md0)';
ls  /target/boot/grub;

echo 'update-grub';
chroot /target /usr/sbin/update-grub;
ls  /target/boot/grub;



in scripts/LAST/60-misc.sh


now it works, it all finishes and boot. The last issue i have is that 
the kernel that boot after this is not capable of findind the root file 
system. It hang up saying 'running local-top /n waiting for root 
filesystem'.


 Both the fai server and the test install server are dell r300 using 
the same kernel so i am puzzled here.


i am now trying to build a kernel with raid included in core to see if 
this will work. I searched but i have not found a way to load the 
modules, adding them to /etc/modules does not seems to do it. I just 
finished it does not change a thing. still the same message even with a 
kernel that include the raid drives in the image.


anything special about the initramsfs that i should do ?



regards,
Jean.

site down

[Jean Spirat]

is it me or is the FAI site down ?

regards,
Jean.

Re: [EMAIL PROTECTED]: Re: nfs headache: can't read superblock]

[Jean Spirat]

Most probably you've been hit by the issue reported here:

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=495535

But you seem to have worked around it anyway :-)

Best,
Michael

  
that seems to be the case, anyway as i know what fai-chboot does now i 
can go and do it by hand.


thanks for pointing it out :)

Jean

fai-chboot (was: nfs headache)

[Jean Spirat]

ok,

using
fai-chboot -v  -Is  2.6.26.3-vs2.3.0.35.5  demo1


gived me:

append initrd=initrd.img-2.6.26.3-vs2.3.0.35.5 ip=dhcp  root=/dev/nfs 
nfsroot=/srv/fai/nfsroot boot=live   FAI_ACTION=install


so the simpler seems the most efficient !



fai-chboot  -f "verbose,sshd,reboot,VSERVERHOST,SETUPSTORAGE"  -k 
"ip=dhcp initrd=initrd.img-2.6.26.3-vs2.3.0.35.5 root=/dev/nfs 
nfsroot=192.168.1.72:/srv/fai/nfsroot/live/filesystem.dir"
vmlinuz-2.6.26.3-vs2.3.0.35.5  demo1


then pxelinux do not changed his config so i added:

append ip=dhcp initrd=initrd.img-2.6.26.3-vs2.3.0.35.5 root=/dev/nfs 
nfsroot=192.168.1.72:/srv/fai/nfsroot/live/filesystem.dir



and there it worked (in fact no but at least it goes to more step and 
stop  asking me the runlevel...).


is this normal that the pxe config is not updated by the  append line 
? Perhaps i do something wrong here ?



JEan

Re: [EMAIL PROTECTED]: Re: nfs headache: can't read superblock]

[Jean Spirat]

Michael Tautschnig a écrit :

I've got no idea, maybe nfs mount from localhost just doesn't work?

Best,
Michael

PS.: This doesn't seem to be a problem in FAI itself...

  


yes this is not related to FAI directly but only inderctly :)

for an fai related one, i just had an issue with fai-chboot where i tried :

fai-chboot  -f "verbose,sshd,reboot,VSERVERHOST,SETUPSTORAGE"  -k 
"ip=dhcp initrd=initrd.img-2.6.26.3-vs2.3.0.35.5 root=/dev/nfs 
nfsroot=192.168.1.72:/srv/fai/nfsroot/live/filesystem.dir"
vmlinuz-2.6.26.3-vs2.3.0.35.5  demo1


then pxelinux do not changed his config so i added:

append ip=dhcp initrd=initrd.img-2.6.26.3-vs2.3.0.35.5 root=/dev/nfs 
nfsroot=192.168.1.72:/srv/fai/nfsroot/live/filesystem.dir



and there it worked (in fact no but at least it goes to more step and 
stop  asking me the runlevel...).


is this normal that the pxe config is not updated by the  append line ? 
Perhaps i do something wrong here ?



JEan

nfs headache: can't read superblock

[Jean Spirat]

Hi,

i am trying to make FAI work on my debian box but i fail at the 'mount 
root fs' part on NFS:


fai02:/# rpcinfo -p 192.168.1.72
  program no_version protocole  no_port
   102   tcp111  portmapper
   102   udp111  portmapper
   132   udp   2049  nfs
   132   tcp   2049  nfs
   151   udp942  mountd
   152   udp942  mountd
   151   tcp945  mountd
   152   tcp945  mountd
   1000241   udp  48209  status
   1000241   tcp  39959  status

fai02:/# mount -v -t nfs 192.168.1.72:/srv/fai/config /mnt/nfs/
mount: 192.168.1.72:/srv/fai/config: can't read superblock


 For me NFS seems to run ok but if i try to mount via nfs i got this 
strange error about superblock, any idea ?


regards,
Jean.

nfs-kernel-server and nfs-user-server

[Jean Spirat]

hi,

is there a way to use the nfs-user-server with FAI (debian etch) and 
not the kernel one ?   The debian package seems to prevent that and the 
kernel i use in the fai server is not compatible with it.


 anybody could help ?

regards,
Jean.

Re: new to the list and fai

[Jean Spirat]

I don't know what kind of things might be necessary with AOE.



having aoe in the kernel does not seems to be enough. The kernel do not 
find the aoe device i think it need an aoe-discover for that :((


But, anyway, I guess you should try it like this, and see what happens 
- report back here or ask on irc if it doesn't work directly.


Another things you might consider, if you use AOE anyway to mount the 
root FS for running the systems after installation, and probably boot 
the systems from PXE anyway, you might consider to build your initial 
installation from a central server, just mount the root fs there, and 
do a FAI dirinstall. 

yes i can, but i also have other setup with disk so i want both :)

the command i used is :

fai-chboot -v  -k 'aoe_iflist="eth1" root=/dev/etherd/e10.1 
rootfstype=ext3' '2.6.22.19-vs2.2.0.7' demohost2

Booting kernel 2.6.22.19-vs2.2.0.7
append   aoe_iflist="eth1" root=/dev/etherd/e10.1 rootfstype=ext3

Kernel parameters: aoe_iflist="eth1" root=/dev/etherd/e10.1 rootfstype=ext3
demohost2 has 192.168.1.202 in hex C0A801CA
Writing file /srv/tftp/fai/pxelinux.cfg/C0A801CA for demohost2



the kernel boot and do a kernel panic because it cannot find the root 
device, but if i grep ALL the files for "aoe_iflist" none has it so 
where does fai store this part ?


The commands
find /srv -exec egrep -l aoe_iflist {} \;
find /etc -exec egrep -l aoe_iflist {} \;
return no result, are those options just sent in the wind or do i miss 
something :) ?


also if i read the fai-chboot man page it give this exemple:
fai-chboot -v atom-localboot /dev/hda1 atom03

for me it is:

fai-chboot -v '2.6.22.19-vs2.2.0.7' '/dev/etherd/e10.1' demohost1


but then i got this error:fai-chboot: unknown host: 
/dev/etherd/e10.1


strange no ?

the issue is that nfs-kernel-server is incompatible with the kernel i 
use on the fai-server so i try to find another way for me, 
nfs-user-server cannot be installed the debian fai package prevents it 
completly (why ?)  and i do not know how to force a package 
install/uninstall withtout looking at dependency in debian.


i go back to googling.


Jean

Re: new to the list and fai

[Jean Spirat]

after a little kernel info reading it seems i just have to replace the 
/dev/nfs by root=/dev/eth10.5 rootfstype=ext3 and it should work ? How 
can i setupt this in FAI ?



is it something like:

fai-chboot -v aoe-enabled-kernel  /dev/eth10.5  testhost


?

regards,
Jean

new to the list and fai

[Jean Spirat]

hello list,

 i am new to the list and trying to use FAI for a fancy setup i am 
building :) 

 I will use an AOE storage blade. This mean that if the kernel is AOE 
aware there should be no need for nfs or things like this. I wondered if 
anyone has any experiences on this ?


 So my question is , can we build a special kernel for FAI that will be 
AOE enabled and make it so that FAI loads the filesystem image using a 
local mount instead of nfs ?


 I think i know how to change the kernel used by FAI to boot, now i 
need to replace nfs with an etherdrive mount of /dev/eth10.5 :)



regards,
Jean.