This topic contains 6 replies, has 0 voices, and was last updated by 
-
Posts
-
Anyone know of a good howto or information to get booting from a hard-drive?
No I don’t mean the gzip -d file.gz > /dev/sdX
The above method still acts a lot like a cd-rom except your booting the image from the hard-drive. The problem is a lot of directories are still read-only which is what I need to avoid. To get systems onboard ethernet drivers working I had to recompile a newer version of the e1000e network driver but there is no way to replace the original as that part of the image is read-only. I need to install it onto a disk where the entire system will have full read-write access (as it should being installed on an HD).
/Databasedirectory is mounted RW. You may store there everything you want and then use preboot or postboot scripts to copy them after a reboot.
@ppalias wrote:
/Databasedirectory is mounted RW. You may store there everything you want and then use preboot or postboot scripts to copy them after a reboot.
The problem is that the weird device names that zeroshell uses doesn’t work very well when I have to rmmod and then insmod the ethernet drivers. I think I got enough of a hint from:
http://www.zeroshell.net/eng/forum/viewtopic.php?t=210
to do what I want. At this point I am getting a kernel panic at boot (VFS can’t find root) but I believe this can easily be solved by copying the ahci and ata_piix modules (my chipset should use one of those) to the initrd.gz image file.
So the initrd was giving me problems so I compiled a new (2.6.30). Its booting and detects my NICs but they are eth0,eth1,eth2,eth3, etc… instead of ETH00, ETH01, ETH02, ETH03.
Do you know how they get the ETHXX names? Is it a modification to the kernel source or does something call ifrename?
I created an /etc/iftab but even after renaming the interfaces and making entries in /var/register zeroshell was saying ‘hardware removed’ on the network configuration page.
I think I tracked down the issue to sysfs not being mounted. After I edited the hardware detection script which runs a for loop to load modules and after that everything appears to be working as expected from the web-interface. Also interesting enough the dev order is the same of what my iftab was even though I removed the ifrename command in the hardware detection script. I will be testing it out a bit more today.
The only issue left I have (that I know of) is that when I connect to SSH the text is all mesed up and I get an error about the terminal. This is also preventing me from logging in via SSH. I get the same type of output even when doing ssh localhost from the zeroshell box itself. Here is what it looks like:
Anyone have any ideas what could cause this?
Terminal emulation VT100?
@ppalias wrote:
Terminal emulation VT100?
This was another problem with mounts. Aparrantly I needed to mount /dev/pts which fixed the issue now my machine is working properly.
I also edited the /etc/passwd to not use the zeroshell front end script and added a command for that instead (for easyness to remember). Not being able to scp anything to the machine made things quite annoying. Now everything appears to be working correctly:
root@zeroshell root> uname -a
Linux zeroshell.example.com 2.6.30 #2 SMP Fri Jun 11 18:49:33 PDT 2010 i686 GenuineIntel unknown GNU/Linux
root@zeroshell root> df -h
Filesystem Size Used Avail Use% Mounted on
rootfs 30G 1.5G 27G 6% /
/dev/root 30G 1.5G 27G 6% /
shm 1.5G 0 1.5G 0% /dev/shm
root@zeroshell root>I am thinking about going to a 64-bit kernel though (mainly so byte counts and stuff dont reset @ 4 GB) and proper atom CPU support (maybe I was just blind but I didn’t see atom listed as an option when compiling the kernel)).
You must be logged in to reply to this topic.