Hello all!

Another day, another question! I am attempting to boot a diskless client using a reduced ramdisk. I would post my notes here, but I don't think anyone's going to want to read that much! I have posted a screenshot of the error it gives below.

Anyone have any idea how I repair this issue? The problem seems to be fairly clear: that it is looking for /etc/profile but can't find it. Assuming this is correct, I'm at a loss for how to correct it.

If you're interested in this problem please let me know, I have taken exhaustive notes and would be willing to share them with whoever would be interested.

Thanks for taking the time! Much appreciated.

-Steve

Attached Thumbnails

 

"a diskless client using a reduced ramdisk"

Where did you get this from? Did it ever work? Did you use some web page for example? Did this client ever work properly on pxe boot?

Hey there jefro, nice to hear from ya' again!

Yes, I've tried to use two different howtos to get this to work... that may even be some of my problem.


The first one I tried was from server-world.info
// This one is for the PXE Server setup
http://www.server-world.info/en/note...OS_6&p=pxe&f=1
// This one is for setting up to boot a diskless client
http://www.server-world.info/en/note...OS_6&p=pxe&f=1

The second one I tried was a much more involved one I found here
http://eduardo-lago.blogspot.com/201...-diskless.html


One problem I am seeing is that the "dracut-network" install from the first howto seem to be having a hand in this boot process even when I changed everything (but did not remove dracut) to the second howto. However, the error seems to point to that it is a problem with my init script (from the 2nd howto) which I don't really understand how to build well enough to 4-wheel-drive it off into the sunset (ie: I can't modify the script).

The beginning on the script looks something like this:
#!/bin/sh

# Make all core utils reachable
. /etc/profile

Which is where the load appears to hang, but being as I'm new to diskless PXE booting, I don't really even know how to debug the process. My successes in PXE have been with booting installs. Anyways, here is how the /etc/profile was created:
$ dd of=etc/profile << EOT
>export PATH="/bin:/sbin:/usr/bin:/usr/sbin:/usr/local/sbin"
>EOT

In the end, the init script was "compiled" using the command:
$ sudo find . -print0 | sudo cpio --null -ov --format=newc | gzip -9 > ../initramfs.cpio.gz

... which I am a bit comfused as to the function of this...

Anyways thanks for the response!.. cheers to you, buddy! If you are interested in helping me out with this, just let me know what information you'd like to see, it's quite a bit of contrasting information between the two howtos and my own exhaustive notes! Have a nice one!

Something that got glossed over is how you made this init file. I suspect this is the problem. You may not have made or have enough ramdisk to use in this also. Without a test of this or knowing what you have I can't even guess.

Basically if you did boot it from a tftp server that is partly a good sign. You'd need to tell us how much ram you have and how old this client is.


Some older pxe nic's stink. I tried to make a 486 with 24M boot pxe and took me a while and it was slow. Actually I used gpxe since I didn't have a way to flash the rom on the nic.

There are 1000 or more web pages on how to boot pxe and not one is the same.

For grins, maybe get a www.netboot.me gpxe floppy and see if this client will boot to web.

I always try to suggest to people that they attempt Knoppix terminal server. It is a set of scripts to take a live cd booted knoppix and allow you to pxe boot a lan client to knoppix or freenx of knoppix. There you can see how the scripts set up tftp and dhcp and how pxelinux and config file 0 is set up.

I don't mean to hog everyone else in their help either.

Some awesome ideas there, buddy! Sorry for not posting all of the configs... I didn't want you to run away on the amount of content I put in the post... but since you asked, here is how I made the init file. These are basically the same steps on:

http://eduardo-lago.blogspot.com/201...-diskless.html

But they are my exact notes since I didn't follow the howto exactly, but pretty close:

**********************************************************************

//Create a compressed filesystem....
// Create a work zone

[root@sdserver cent]# cd
[root@sdserver ~]# mkdir -p /tmp/cache
[root@sdserver ~]# mkdir /tmp/wrk
[root@sdserver ~]# pushd /tmp/wrk
/tmp/wrk ~

// Create and enter initramfs root directory

[root@sdserver wrk]# mkdir initramfs
[root@sdserver wrk]# ls
initramfs
[root@sdserver wrk]# pwd
/tmp/wrk
[root@sdserver wrk]# pushd initramfs/
/tmp/wrk/initramfs /tmp/wrk ~
[root@sdserver initramfs]# pwd
/tmp/wrk/initramfs

// Create filesystem base directories

[root@sdserver initramfs]# mkdir -p -m 0755 dev etc/{,init,sysconfig} mnt sys usr/{,local} var/{,www,log,lib,cache} run
[root@sdserver initramfs]# mkdir -p -m 0555 {,s}bin lib{,32,64} proc usr/{,s}bin
[root@sdserver initramfs]# mkdir -p -m 0700 root
[root@sdserver initramfs]# mkdir -p -m 1777 tmp
[root@sdserver initramfs]# pushd var
/tmp/wrk/initramfs/var /tmp/wrk/initramfs /tmp/wrk ~
[root@sdserver var]# ln -s ../run run
[root@sdserver var]# popd
/tmp/wrk/initramfs /tmp/wrk ~

// Create /etc/profile to exports environ. variables

[root@sdserver initramfs]# dd of=/etc/profile << EOT
> export PATH="/bin:/sbin:/usr/bin:/usr/sbin:/usr/local/sbin
> EOT
0+1 records in
0+1 records out
59 bytes (59 B) copied, 0.000161676 s, 365 kB/s

// Create /etc/fstab with mount points

[root@sdserver initramfs]# dd of=etc/fstab << EOT
> devpts /dev/pts devpts nosuid,noexec,gid=5,mode=0620 0 0
> tmpfs /dev/shm tmpfs nosuid,nodev,mode=0755 0 0
> sysfs /sys sysfs nosuid,nodev,noexec 0 0
> proc /proc proc nosuid,nodev,noexec 0 0
> EOT
0+1 records in
0+1 records out
303 bytes (303 B) copied, 6.0199e-05 s, 5.0 MB/s

[root@sdserver initramfs]# vi /etc/fstab
/dev/mapper/vg_sdserver-lv_root / ext4 defaults 1 1
UUID=eb47573a-3ae0-4854-a13b-af4c091e3fd4 /boot ext4 defaults 1 2
/dev/mapper/vg_sdserver-lv_home /home ext4 defaults 1 2
/dev/mapper/vg_sdserver-lv_swap swap swap defaults 0 0
tmpfs /dev/shm tmpfs defaults 0 0
devpts /dev/pts devpts gid=5,mode=620 0 0
sysfs /sys sysfs defaults 0 0
proc /proc proc defaults 0 0

[root@sdserver initramfs]# chmod 0644 etc/fstab

// Config passwd & group settings

[root@sdserver initramfs]# dd of=etc/passwd << EOT
> root:x:0:0:root:/root:/bin/sh
> nobody:x:99:99:NoBody:/none:/bin/false
> www:x:33:33:HTTP Server:/var/www:/bin/false
> EOT
0+1 records in
0+1 records out
113 bytes (113 B) copied, 5.8441e-05 s, 1.9 MB/s

[root@sdserver initramfs]# dd of=etc/group << EOT
> root:x:0:
> nobody:x:99:
> www:x:33:
> EOT
0+1 records in
0+1 records out
33 bytes (33 B) copied, 6.7918e-05 s, 486 kB/s

// Configure some host related settings:

[root@sdserver initramfs]# dd of=etc/host.conf <<< "multi on"
0+1 records in
0+1 records out
9 bytes (9 B) copied, 5.3581e-05 s, 168 kB/s

[root@sdserver initramfs]# dd of=etc/hostname <<< "rambox"
0+1 records in
0+1 records out
7 bytes (7 B) copied, 5.5115e-05 s, 127 kB/s

[root@sdserver initramfs]# dd of=etc/hosts << EOT
> 127.0.0.1 localhost.localdomain localhost
> 127.0.0.1 $(cat etc/hostname)
> EOT
0+1 records in
0+1 records out
59 bytes (59 B) copied, 6.125e-05 s, 963 kB/s

// Configure timezone:

[root@sdserver initramfs]# dd of=etc/timezone <<< "America/Phoenix"
0+1 records in
0+1 records out
16 bytes (16 B) copied, 5.2888e-05 s, 303 kB/s

[root@sdserver initramfs]# cp /usr/share/zoneinfo/$(cat etc/timezone) etc/localtime

// Getting busybox and create sh symbolic link:

[root@sdserver initramfs]# pushd bin
[root@sdserver bin]# dirs -v
0 /tmp/wrk/initramfs/bin
1 /tmp/wrk/initramfs
2 /tmp/wrk
3 ~

[root@sdserver bin]# chmod +w .
[root@sdserver bin]# bb=busybox-x86_64 && wget $bb http://www.busybox.net/downloads/bin...busybox-x86_64
--2013-07-03 00:10:33-- http://busybox-x86_64/
Resolving busybox-x86_64... failed: Name or service not known.
wget: unable to resolve host address “busybox-x86_64”
--2013-07-03 00:10:33-- http://www.busybox.net/downloads/bin...busybox-x86_64
Resolving www.busybox.net... 140.211.167.224
Connecting to www.busybox.net|140.211.167.224|:80... connected.
HTTP request sent, awaiting response... 200 OK
Length: 1089864 (1.0M) [text/plain]
Saving to: “busybox-x86_64”

100%[======================================>] 1,089,864 648K/s in 1.6s

2013-07-03 00:10:35 (648 KB/s) - “busybox-x86_64” saved [1089864/1089864]

FINISHED --2013-07-03 00:10:35--
Downloaded: 1 files, 1.0M in 1.6s (648 KB/s)

[root@sdserver bin]# cp busybox-x86_64 busybox && chmod +x busybox
[root@sdserver bin]# cp busybox /tmp/cache/busybox && chmod +x /tmp/cache/busybox
[root@sdserver bin]# ln -s busybox sh
[root@sdserver bin]# chmod -w .
[root@sdserver bin]# dirs -v
0 /tmp/wrk/initramfs/bin
1 /tmp/wrk/initramfs
2 /tmp/wrk
3 ~
[root@sdserver bin]# popd
/tmp/wrk/initramfs /tmp/wrk ~

// create an script named renew_ip as our DHCP script

[root@sdserver initramfs]# pushd bin
/tmp/wrk/initramfs/bin /tmp/wrk/initramfs /tmp/wrk ~
[root@sdserver bin]# chmod +w .
[root@sdserver bin]# wget http://git.busybox.net/busybox/plain.../simple.script
[root@sdserver bin]# cp simple.script /tmp/cache/simple.script
[root@sdserver bin]# chmod +x simple.script
[root@sdserver bin]# dd of=renew_ip << EOT
> #!/bin/sh
>
> ifconfig eth0 up
[root@sdserver bin]# dd of=renew_ip << EOT
> udhcpc -t 5 -q -s /bin/simple.script
> EOT
0+1 records in
0+1 records out
37 bytes (37 B) copied, 9.3154e-05 s, 397 kB/s

[root@sdserver bin]# chmod +x renew_ip
[root@sdserver bin]# chmod -w .
[root@sdserver bin]# popd
/tmp/wrk/initramfs /tmp/wrk ~

// launching the shell or by executing our own /sbin/init w/o remounting the
// root (/)

[root@sdserver initramfs]# vi init
#!/bin/sh

. /etc/profile
/bin/busybox --install -s
mkdir dev/pts
mkdir dev/shm
chmod 1777 dev/shm
mount -av
ln -sf /proc/mounts /etc/mtab
touch /etc/mdev.conf
echo /sbin/mdev > /proc/sys/kernel/hotplug
mdev -s
hostname=$(cat /etc/hostname)
shell() {
echo "${1}Launching shell..." && exec /bin/sh
}
grep -q 'shell' /proc/cmdline && shell
for i in $(cat /proc/cmdline); do
par=$(echo $i | cut -d "=" -f 1)
val=$(echo $i | cut -d "=" -f 2)
case $par in
root)
root=$val
;;
init)
init=$val
;;
esac
done
init=${init:-/sbin/init}
root=${root:-/dev/hda1}
if grep -q 'rambox' /proc/cmdline ; then
[ -e ${init} ] || shell "Not init found on ramdisk at '${init}'... "

echo "Keeping the ramdisk since rambox param was supplied & executing init... "
exec ${init}

#This will only be run if the exec above failed
shell "Failed keeping the ramdisk and executing '${init}'... "
fi
mkdir /newroot
mount ${root} /newroot || shell "An error ocurred mounting '${root}' at /newroot... "
[ -e /newroot${init} ] || shell "Not init found at '${init}'... "

echo "Resetting kernel hotplugging... "
:> /proc/sys/kernel/hotplug
echo "Umounting all... "
umount -a
echo "Switching to the new root and executing init... "
exec switch_root /newroot ${init}
mount -av
mdev -s
shell "Failed to switch_root... "

[root@sdserver initramfs]# chmod +x init
[root@sdserver initramfs]# chown -R root:root *
[root@sdserver initramfs]# find . -print0 | sudo cpio --null -ov --format=newc | gzip -9 > ../initramfs.cpio.gz

[root@sdserver initramfs]# cp ../initramfs.cpio.gz /var/lib/tftpboot
[root@sdserver initramfs]# dirs -v
0 /tmp/wrk/initramfs
1 /tmp/wrk
2 ~
[root@sdserver initramfs]# popd
/tmp/wrk ~
************************************************************************

There is more to it, but the error seems to point to something here in how I made this init script.

As for hardware, the old machine actually the server, the client I'm trying to boot is actually fairly new. The server is an HP ProLiant ml310 g4 (circa 2007 or something... got it at a community college liquidation for 20$!). The client I'm trying to boot is an Asus laptop w/ an Intel i3 and 4GB RAM.

I have managed to get it to successfully catch a PXE install, and then install to an iSCSI device, but never to get it to actually boot the operating system from the same iSCSI. I've got the linux bug though and I can't stop experimenting! PXE booting an actual OS seems to be my kryptonite at the moment!

Thanks again for the time, jefro, much appreciated!

Hope I get time to look at this later.

Thanks, buddy, I appreciate all your help. I'll be spending most of my today trying to better understand this problem myself. If you don't get a chance to look at it then thanks for your help regardless.

Never saw that dd of=/to/a/file << EOF. Nice way for dd usage.

Nother thing. Your using /bin/sh in your init. sh does stop at the first error it encounters and as the file /etc/profile does not seem to exsist or is not readable sh terminates that terminates the init process which terminates the kernel...

Can you give the dhcp config? Maybe the pxeboot config as well. Not so deep into pxe but used FAI so got some clues.

Can you go to this client, load linux on it and then make a intrafs using dracut? Could be a simple issue going on or one small thing in what you posted but anytime you make this stuff, make it with the most minimum to start. Then add in needed stuff later.

https://wiki.gentoo.org/wiki/Dracut#...nitramfs_Image

https://fedoraproject.org/wiki/Dracut

One problem with making an image on a different system is that usually only the hosts devices are being sent to the client image.

I'd still verify that you can pxe boot or gpxe boot.

zhjim:
Thanks, I thought so, too. This was actually my first time using dd in this manner, but ironically I was already trying to learn how to use << EOT when I found this dd command usage. Here is the DHCP config I used:

# dhcpd.conf
#
# DHCP configuration file for ISC dhcpd
#

# Use this to enble / disable dynamic dns updates globally.
ddns-update-style none;

# Definition of PXE-specific options
# Code 1: Multicast IP address of boot file server
# Code 2: UDP port that client should monitor for MTFTP responses
# Code 3: UDP port that MTFTP servers are using to listen for MTFTP requests
# Code 4: Number of seconds a client must listen for activity before trying
# to start a new MTFTP transfer
# Code 5: Number of seconds a client must listen before trying to restart
# a MTFTP transfer
option space PXE;
option PXE.mtftp-ip code 1 = ip-address;
option PXE.mtftp-cport code 2 = unsigned integer 16;
option PXE.mtftp-sport code 3 = unsigned integer 16;
option PXE.mtftp-tmout code 4 = unsigned integer 8;
option PXE.mtftp-delay code 5 = unsigned integer 8;
option PXE.discovery-control code 6 = unsigned integer 8;
option PXE.discovery-mcast-addr code 7 = ip-address;

subnet 192.168.1.0 netmask 255.255.255.0 {

class "pxeclients" {
match if substring (option vendor-class-identifier, 0, 9) = "PXEClient";
option vendor-class-identifier "PXEClient";
vendor-option-space PXE;

# At least one of the vendor-specific PXE options must be set in
# order for the client boot ROMs to realize that we are a PXE-compliant
# server. We set the MCAST IP address to 0.0.0.0 to tell the boot ROM
# that we can't provide multicast TFTP (address 0.0.0.0 means no
# address).
option PXE.mtftp-ip 0.0.0.0;

# This is the name of the file the boot ROMs should download.
filename "pxelinux.0";

# This is the name of the server they should get it from.
next-server 192.168.1.5;
}

pool {
max-lease-time 86400;
default-lease-time 86400;
range 192.168.1.50 192.168.1.100;
allow unknown clients;
}

host rambox {
hardware ethernet 14:da:e9:af:02:f3;
fixed-address 192.168.1.201;
hostname rambox.sddomain.local;
}

}

***************************************************************
And here is the PXE config:

// Configure PXE environment:

[root@sdserver cent]# cp /usr/share/syslinux/pxelinux.0 /var/lib/tftpboot

The conversion result:
IP dotted quad IP decimal IP hex IP Binary
192.168.1.201 3232235977 C0A801C9 11000000 10101000
00000001 11001001

// The Table doesn't show on properly on Linux Questions
// The idea above is that 192.168.1.201 translates to C0A801C9 in hex

[root@sdserver cent]# vi /var/lib/tftpboot/pxelinux.cfg/C0A801C9
DEFAULT bzImage
APPEND initrd=initramfs.cpio.gz rw ip=192.168.1.201:192.168.1.5:192.168.1.1:255.255.255.0:rambox:eth0ff shell

The reasoning behind using the IP hex naming I found in this little tidbit:
"The Linux PXE boot loader uses its own IP address in hexadecimal format to look for a single configuration file under pxelinux.cfg directory, if its not found it will remove the last octet and try again, repeating until it runs out of octets."

Jefro:
Thanks for the ideas. I hadn't really thought of the fact that making my image on a separate machine might cause a problem. I will see if I can't make my initramfs on the client machine, and I'll take a look through the links you posted & respond to them in a bit.

As for the client machine itself, it is capable of PXE boot... one of my coworkers (the one who gave me this project) showed me in typical genius fashion: he did it real quick without much explanation, then let me run on trying to figure out how he did it. And he's not interested in handing out the answers, so I'm to figure it out for myself anyway possible.

Anyways, the client is capable of booting over the network... the real question is: am I capable of booting it over the network!

Thanks for the response, always a pleasure to hear from ya', and thanks for the direction.

After poking around the configs I made in this project, I see that files like "/etc/profile" were modified but not copied into /var/lib/tftboot/ anywhere. Could this be where my problem is?

That was a first error. Guess it could have been the issue.


The next line concerned me more.

https://www.linuxquestions.org/quest...l-init-353920/

So, if I understand the link you sent correctly, the "/init: .: line 3: can't open '/etc/profile'" is not the actual error I'm looking for? Instead my problem is in the "Pid: 1, comm: init Not tainted 2.6.34.14-minimal #1"?

My apologies if I misunderstood what you were getting at, but that was a very interesting link you sent, thanks. Any ideas on how I might go about fixing it? And thanks for the reply, I appreciate it.

I still think that the init process dies cause of not able to get to /etc/profile.

Anyways lets break the whole process in pieces. PXE is just shovels your ramdisk to the client over the network. Which works okay. It's just that the init process of the ramdisk is just not happy. So you need to get your init script sorted inside your ramdisk. So we can ignore PXE for the time beeing and concentrate on the ramdisk. Can you boot this inside a virtual machine or directly from a client? So it might be easier to test it. Just like an extra grub entry.

Or maybe after creating the ramdisk unpack it inside /mnt or similar. Maybe it can also be mounted through a loop device?