Im relatively new to the OpenWrt scene and have only started to realize it's potential.  To get my feet wet, I bought three V.3 WRT54G's.  I lodaed the jffs2 bin and it booted up fine.  I am not used to the Telnet/SSH commands.  I have putty and WinScp3, but dont know how to load a package.  I read through the docs and forums and found a lot of info but the commands...  is it wget? filename? directory?  Also are there custom web interfaces?  I am going to continue to read through the site.  I just feel like a kid with a new toy without the batteries.

- first you have to connect via telnet or ssh to the router

command: telnet 192.168.1.1

- second you have to edit the file /etc/ipkg.conf

in this file you should have an url which is reachable, so if you can't connect to the internet you need a local webserver with the package direcotry

editing: vi /etc/ipkg.conf

-third you run following command:

ipkg update

now you can install a package by running following command on the router:

ipkg install "Packagename"

-dryades

Thank you sir!  I was able to update the ipkg.  My next step will be to get a web interface...I haven't seen any for this build.

I too just flashed my new v3 router with OpenWRT the experimental build with jffs2.  It worked flawlessly.  I could not do the boot_wait instructions as the tftp server did not accept the firmware from my client.  I updated ipkg manager and have loaded some packages.  I am still trying to learn how to set everything. 

If I run nvram commit, does it write all my current settings to flash?  If so, does this mean I don't have to write scripts to set everything at boot?  I guess it makes sense.  Read most of the docs, but they seem vague in places.  For example, how do you tell which ethernet port a pc is plugged into from the console?  Just want to understand it all.  I'm learning as I go, but do have a very good linux background.  Like it so far.

With the exception of /tmp (which is a ramdisk), all the writable areas of the filesystem are stored directly on flash using the jffs2 filesystem.

Instead of config files, the linksys firmware included one writable section on the flash which stored nvram variables, simple "name=value" combinations. The command "nvram show" lists all nvram variables currently set (cached in memory) and nvram commit writes those variables to flash. Without an nvram commit, changes are in memory only and will be lost on reboot.

Why do we still use nvram when we have a writable filesystem? Part of it is legacy, it allows people to quickly get openwrt up and running with the same settings they had previously, but the main reason is that we're required to have an nvram partition* and since it already holds the settings it seemed redundant to waste space storing the same settings elsewhere.

Why don't we store all settings in nvram? There's some firmwares that do that, but just because it can be done doesn't mean it's a good idea. The firmwares that store things like ssh keys in nvram make a mess out of the nvram show output and require the key to be moved to the filesystem (a ram drive) before it's used anyways.

--

* Various portions of the bootstrap that loads the kernel, the kernel itself, modules and even some utils require nvram.

(feel free to update the wiki with the above)

Thanks for the post.  I get the idea, just getting used to the syntax of everything and trying to find it all.  I work on Cisco hardware all the time, so this is very normal.  You could compare this to your running config versus what is in nvram (saved config).  nvram commit = write mem.