Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Sheevaplug - pure Linux geek juice!!!

4 views
Skip to first unread message

7

unread,
Aug 22, 2009, 4:40:37 PM8/22/09
to
Sheevaplug - pure Linux geek juice!!!
-------------------------------------

http://www.marvell.com/products/embedded_processors/developer/kirkwood/sheevaplug.jsp

Not very often you get a high end distro like Ubuntu 9.04
and a cheap embedded Linux developer system that is also
capable of running a high end distro.

The $99 Sheevaplug is very much it.

You get a main adapter size box with a plug to
plug into mains. It has ethernet link, SDCard, USB
and a mini USB serial port.

Nothing else.

No monitor port, no keyboard, no mouse port, no nothing.
Its also running on an ARM chip!

So what does it do except sit there and do nothing?

Welll..... its the ultimate food from the
gods sent to geeks!!!

Since it has 'nothing' you have to bring it alive yourself
and mould it to your whims!!!


It is pure geek wallowing in Linux from here on in...


First connect the mini USB serial port to your
Linux Ubuntu desktop USB port. Running dmesg in a console
window shows it has been recognised and installed as /dev/ttyUSB1
So you need some software to talk to the Sheevaplug.
Install minicom with
apt-get install minicom
in a console window.

Then run
minicom -s
to and change set up for 115200 baud 8 bits, N for no parity, 1 stop bit.
Change the default serial device to /dev/ttyUSB1
Press return to save changes.
Press save changes as default
Then Exit minicom.
Then run
minicom
again, and now you are connecting OK to the Sheeva plug through the
serial port.

Login as root with password nosoup4u
(yup that is the default password!)

First off change the password to something you like with the
passwd
command.

You can issue command like
shutdown -r now
and see the computer shutting down and rebooting like a real
computer through minicom.

Log back in, connect up the ethernet, give it a few
seconds and then enter
ifconfig
It shows the IP address of the Sheevaplug.

Now you can abandon the minicom console and start another
bash console window and connect via ssh!!!
ssh root@ipaddress_of_sheevaplug
So now you are in and looking around.

If you have konqueror installed on your desktop computer,
you can also get in by opening konqueror and entering in the URL
fish://root@ipaddress_of_sheevaplug
to snoop around to see what other software has been pre-installed.

First thing to do is to get the existing OS of the internal
flash chip (to avoid wearing it down) and run it off an SD Card
formatted with EXT2 to minimise wearing down the SD Card.

The instructions are here
http://plugcomputer.org/plugwiki/index.php/Category:HowTo
(Choose from all the howto available in that document
and don't use EXT3, use EXT2 - EXT3 will wear down your
SD Card. I used a 16Gb SD card and set aside 1Gb for swap
space - excessive because I didn't want to risk running
out of space for later parts of the experiments)


Having transferred everything off the internal flash and booting
off the SDcard allows you the freedom to install
lots of new software.

First off got a few of these installed
apt-get install wget

But it came up with error message - the issue is that two
directories are always in RAM and can get filled - so it forgets
to create them whenever Sheevaplug boots.
So create these two directories to get apt-get working

mkdir /var/cache/apt/archives
mkdir /var/cache/apt/archives/partial

Then I test downloaded a file using wget.
I then copied out the file from the sheevaplug using scp command.
So in effect the sheevaplug has just become my downloader
device consuming 5W that can be left running all night if need be.

and then installed away some of the following

apt-get install lynx
apt-get install rsync
apt-get install youtube-dl
apt-get install sqlite3
apt-get install bash
apt-get install xterm

(If your /var/cache/apt/archives gets full,
then reboot the sheeva plug and recreate the
directories again)

(installing bash might have been redundant - but even
if redundant, you can see how it reacts to the command
to see if updates are available)

then I did

apt-get update
apt-get upgrade

to upgrade the Linux to the latest version (but you may
have to do that before installing anything if some
of the packages got updated).

Then tried to get VNC server working
http://computingplugs.com/index.php/Installing_X11
Got that working.
(Be weary if you follow instructions to recompile from
source in that link to get xterm working
Your dependency checking may go out the window.)

And then got that working!!!!
So even through Sheevaplug has no graphics card,
I can still get a VNC session on it going!!!

Then installed more complex software

apt-get install firefox
apt-get install pan
apt-get install openoffice.org
apt-get install mysql-server
apt-get install qcad
apt-get install konqueror
apt-get install xpdf
apt-get install dillo
apt-get install gnumeric
apt-get install abiword
apt-get install apache2

All in all 2Gb of software got installed.
Your luck will vary as the software is
constantly being updated (and broken)
so wait a day if something doesn't work
and try again. [My current install has problems
starting X and crashes first time but
second time its OK when running applications
like firefox.] Some things like gambas2 and
its libs are not in the arm repositories yet, so
not everything can work just yet anyway.

GUI software like firefox I can use
ssh -X root@ipaddress_of_sheevaplug
and then run
firefox
to get it to run as a remote X session.

Same with pan, open office, konqueror, gnumeric, abiword,
xpdf, qcad and dillo.

Apache server and mysql server is running!
Its turned the sheevaplug into a powerful server.

But also desktop features available too...

I ran pan (newsreader) and change pan to use dillo (webrowser)
instead of firefox and I get near full speed
reading newsgroups and browsing links
even through I am running remote X session.

Running firefox or dillo with remote X session
is also a safer bet - I can move all the dodgy website
browsing to the sheevaplug and change the SD card if some
dodgy infection is suspected.

For $99, the sheeva plug is 1.2GHz ARM with 512Mbytes of
RAM and 512Mbytes of flash and from the above
you can see it runs a full version
of Ubuntu very well!!!


This post is brought to you from the Sheevaplug
running pan through a remote x session
on the main desktop computer :-)

Have fun!

Tony(UK)

unread,
Aug 23, 2009, 1:53:59 AM8/23/09
to
7 wrote:

> Sheevaplug - pure Linux geek juice!!!
> -------------------------------------

snip How-I-can-muddle-minds-with instructions...

Please explain so that newcomers to Linux can understand - exactly why would
they want or need this?

The small mass that occupies my cranial cavity now hurts, having tried to
process the information.

7

unread,
Aug 23, 2009, 5:06:24 AM8/23/09
to
Tony(UK) wrote:

> 7 wrote:
>
>> Sheevaplug - pure Linux geek juice!!!
>> -------------------------------------
> snip How-I-can-muddle-minds-with instructions...
>
> Please explain so that newcomers to Linux can understand - exactly why
> would they want or need this?

If they are Linux geeks and newcomers to embedded Linux then they would find
this immensely absorbing.

> The small mass that occupies my cranial cavity now hurts, having tried to
> process the information.

I found it absorbing because it frees you up to try desktop Linux
in an embedded Linux context.


Posting this from knode running on sheevaplug :-)


Jerry McBride

unread,
Aug 23, 2009, 3:36:43 PM8/23/09
to
Tony(UK) wrote:

>
>
> 7 wrote:
>
>> Sheevaplug - pure Linux geek juice!!!
>> -------------------------------------
> snip How-I-can-muddle-minds-with instructions...
>
> Please explain so that newcomers to Linux can understand - exactly why
> would they want or need this?
>


For me it means that I can offer CHEAP linux appliances for my charity and
church customers... Imagine... a $100.00 server solution... rock solid...
no expensive mickysoft licensing... just plain, bulletproof technology that
will pretty much fit any IT budget!

Means a lot to me.


--

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

From the desk of:
Jerome D. McBride

15:33:39 up 33 days, 19:38, 5 users, load average: 0.10, 0.29, 0.41

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

Homer

unread,
Aug 23, 2009, 7:16:40 PM8/23/09
to
Verily I say unto thee, that Tony(UK) spake thusly:

> Please explain so that newcomers to Linux can understand - exactly
> why would they want or need this?

It's a server ... the size of a plug.

So is your question "why do they need a server" or "why do they need a
server this small"?

The latter is rather obvious: cost, space and energy efficiency.

As for the former, I can provide several examples of what I personally
use my servers for, but essentially they provides many useful services
that run in one's absence (i.e. one can shut down one's power-guzzling
and noisy desktop, then go out/sleep while the quiet and highly energy
efficient server performs a queue of transactions, potentially lasting
hours or even days).

One example is queuing BitTorrent files overnight (off-peak times), so
those downloads don't use up any peak bandwidth allowances.

Another example is automatically transcoding DVD Video discs to Theora
video on an NFS share, to be streamed over the network.

Another example is a combination firewall, content filtering and cache
service (iptables, Privoxy, Squid), which offloads CPU intensive work,
consolidates maintenance, and provides consistent service, between all
systems connected to the network.

I also run my own IMAP server, which fetches and consolidates /all/ my
E-mail from various sources into one location which I control (so it's
immune to NSLs and other third-party data breaches/loss). This means I
have peace of mind, and the simplicity of setting up E-mail clients to
access only /one/ server.

I run my own NNTP server, enabling me to retain archives (ignoring the
X-No-Archive header) of Usenet posts, perform detailed filtering, then
produce interesting stats based on that data.

I run an OpenVPN /client/ on the server, connected to a commercial VPN
service offshore. But as I forward all requests from the local Privoxy
service through the VPN tunnel, this means HTTP requests on the entire
network (using the proxy) go through the tunnel, essentially turning a
client into a service. Again this offloads CPU intensive activity, and
provides consistent service across the entire network.

As already mentioned, I run an NFS service, on a large array of disks,
consolidated into a single logical volume using Linux LVM2. This gives
me a truly vast and dynamically extensible storage space, which can be
utilised by every other system on the network. As this data (and every
-thing else on the server) is automatically backed up on a daily basis
(to an external array), this also offloads the CPU intensive, and time
consuming activity of individually backing up all my other systems, or
at least the data component (I make LVM snapshot images periodically).

I run a local mirror of several fedora repos (both official and third-
party), which enables me to perform updates to all my systems, without
redundantly downloading the same updates multiple times from upstream.
The subsequent update process is therefore extremely fast, as the RPMs
are then only transmitted across the local gigabit network.

I run a Koji buildsystem, that enables me to automatically rebuild RPM
packages from SRPMs (also mirrored in the above), with $RPM_BUILD_OPTS
set for optimal use of the host architecture (e.g. "-mtune=athlon64").
This also allows me to automatically apply (and continually /reapply/)
my own modifications to those packages (e.g. applying patches to SPECs
that re-enable certain build options that were disabled for reasons to
do with bogus software patents). In this way, updates from upstream do
not reverse my custom modifications, and I always have architecturally
optimised binaries (the latter is especially important for VIA C3, and
AMD K6, which both lack the CMOV instruction, and also for certain big
-endian systems like the NSLU2).

This is just a small selection things I do with my servers. It may all
sounds quite overwhelming to the "newcomers" you mention, but the fact
is that once it's set up, using it is trivially simple, and incredibly
useful.

It's also worth noting that achieving the above with Windows, would be
difficult and expensive at best, and in many cases impossible - mostly
due to lack of sources.

--
K.
http://slated.org

.----
| "Seek not the favour of the multitude; it is seldom got by honest
| and lawful means. But seek the testimony of few; and number not
| voices, but weigh them." ~ Immanuel Kant
`----

Fedora release 8 (Werewolf) on sky, running kernel 2.6.26.8-57.fc8
00:16:09 up 87 days, 4:14, 5 users, load average: 4.42, 4.23, 2.76

Jerry McBride

unread,
Aug 23, 2009, 9:13:28 PM8/23/09
to
Homer wrote:

>
>
> Verily I say unto thee, that Tony(UK) spake thusly:
>
>> Please explain so that newcomers to Linux can understand - exactly
>> why would they want or need this?
>
> It's a server ... the size of a plug.
>

Exactly! For a measly $99.00!

> So is your question "why do they need a server" or "why do they need a
> server this small"?
>
> The latter is rather obvious: cost, space and energy efficiency.
>

I'm picturing a power strip with... any number of them! Imagine a 2.3 to 7
watt server solution...

Simply awesome!

> As for the former, I can provide several examples of what I personally
> use my servers for, but essentially they provides many useful services
> that run in one's absence (i.e. one can shut down one's power-guzzling
> and noisy desktop, then go out/sleep while the quiet and highly energy
> efficient server performs a queue of transactions, potentially lasting
> hours or even days).
>
> One example is queuing BitTorrent files overnight (off-peak times), so
> those downloads don't use up any peak bandwidth allowances.
>

7!... share the strategy for this... Sounds very interesting. I do a lot of
downloading with p2p. Any tips would be welcomed.

> Another example is automatically transcoding DVD Video discs to Theora
> video on an NFS share, to be streamed over the network.
>
> Another example is a combination firewall, content filtering and cache
> service (iptables, Privoxy, Squid), which offloads CPU intensive work,
> consolidates maintenance, and provides consistent service, between all
> systems connected to the network.
>
> I also run my own IMAP server, which fetches and consolidates /all/ my
> E-mail from various sources into one location which I control (so it's
> immune to NSLs and other third-party data breaches/loss). This means I
> have peace of mind, and the simplicity of setting up E-mail clients to
> access only /one/ server.
>
> I run my own NNTP server, enabling me to retain archives (ignoring the
> X-No-Archive header) of Usenet posts, perform detailed filtering, then
> produce interesting stats based on that data.
>

This is my first project. I use leafnode and an external usb attached
harddrive for my solution.

> I run an OpenVPN /client/ on the server, connected to a commercial VPN
> service offshore. But as I forward all requests from the local Privoxy
> service through the VPN tunnel, this means HTTP requests on the entire
> network (using the proxy) go through the tunnel, essentially turning a
> client into a service. Again this offloads CPU intensive activity, and
> provides consistent service across the entire network.
>

Excellent.

Cheers.


--

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

From the desk of:
Jerome D. McBride

21:05:36 up 34 days, 1:10, 5 users, load average: 0.10, 0.25, 0.21

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

Homer

unread,
Aug 24, 2009, 2:17:31 PM8/24/09
to
Verily I say unto thee, that Jerry McBride spake thusly:
> Homer wrote:

>> One example is queuing BitTorrent files overnight (off-peak times),
>> so those downloads don't use up any peak bandwidth allowances.
>
> 7!... share the strategy for this... Sounds very interesting.

http://deluge-torrent.org/

Install and run on the server (using screen or VNC, as it's GUI only),
then configure and run the WebUI and Scheduler plugins. After that, just
leave it running permanently, quit VNC/screen, then connect to it from a
client system using any web browser to add torrents. You'll want to make
sure the target directory for saved data is on an NFS share, so you can
access that too from any client system.

I actually prefer Transmission, as it has a proper daemon that can be
run without X. This also has a WebUI and scheduler (v1.5+ only, in the
latter case). It's a bit more work to get going, but uses far less
resources (running as a service):

http://trac.transmissionbt.com/wiki/HeadlessUsage
http://trac.transmissionbt.com/wiki/EditConfigFiles

--
K.
http://slated.org

.----
| "Seek not the favour of the multitude; it is seldom got by honest
| and lawful means. But seek the testimony of few; and number not
| voices, but weigh them." ~ Immanuel Kant
`----

Fedora release 8 (Werewolf) on sky, running kernel 2.6.26.8-57.fc8

19:17:01 up 87 days, 23:15, 5 users, load average: 0.00, 0.03, 0.08

0 new messages