Upstream/Generic Info Gatering Tool [Was: Pastebin command line util]

Martin Meredith mez at ubuntu.com
Thu Nov 2 18:43:11 GMT 2006


On Thu, 2006-11-02 at 17:41 +0000, Vassilis Pandis wrote:

> You might be interested in the Upstream project: https://wiki.ubuntu.com/Upstream

Seems like an interesting project, and an interesting concept.

I talked on irc with pitti for a little, about possibly setting up a
package for the pastebin script at
http://www.ubuntulinux.nl/files/pastebin

The main reason behind my asking about this is to support users in
IRC/mailing lists with greater ease, for example, using xchat, it would
be simpler to say 

"type /exec -o cat /etc/X11/Xorg.conf | pastebin"

than trying to explain how to pastebin a file, etc etc

There are some issues with using a pastebin script, espescially if it
would be a "default" script to be used by helpers, mainly being the fact
that because ubuntu is supported for up to 10 years, the server-side
pastebin scripts would need to stay constant during that time, or the
client-side scripts would need to go through an SRU each time that the
script was updated in one way or another.

The (easiest) solution to this would be to create an "official" (I use
the term in the loosest sense) pastebin server for ubuntu, which stayed
stable for the amount of time required.

However, looking at the Upstream wiki page, it provides in my humble
opinion, an excellent concept.

Would it really be that hard to create an application (or suite of
applications) that provided info out to the outside world regarding the
most common problems that we face in ubuntu (Video issues, etc etc etc)

I envision that we could easily create a suite of applications (both
client and server-side) that would allow for people to support ubuntu a
lot easier.

Lets give an example use case.

In IRC...

>> user has joined #ubuntu
<user> Hi there! I seem to be having problems with my Video card, my
monitor is stuck at a very low resolution
<helper> Hi there user, welcome to #ubuntu, this is a common problem,
and we need a little more information to help you out
<helper> There's a nice application in ubuntu that will give us all the
information about your system that we need, all you have to do is run
gigt --vid from a console, and paste the URL you recieve back here
<user> thankyou, the URL is
http://gigt.ubuntu.com/3619e096286312354317a68dda869e80
<helper> thankyou, give me a moment while I go and have a look at your
system info

 .... helper goes and looks at the website, finds the problem, and sends
instructions to the user on how to fix this problem.

Result = one happy user, and one helper available for the next issue in
a quicker time.

Obviously, of course, we need to identify the commonest issues in
certain packages (hence copying this to -nun - who will be able to
compile a list regarding the commonest problems)

However, here we again have the issue of stability. However, if we were
to custom code the server (or the scripts on the server hosted by
apache) then we would have complete control over "backwards
compatibilty" meaning that we'd be able to allow it to work concurrently
with all versions of ubuntu.

A thought would be for the client to connect to the server, and then
recieve a "recipe" for what it needed to send to the server (a file
containing for example, a list of files to upload, a list of commands to
capture the output from, etc, etc)

A typical client-server (whether stateless or stateful) would be similar
to

--> Client connects to server 
--> Client sends release of ubuntu and requested recipe (both needed as
recipes can change from release to release)
<-- Server Verifies the release is valid (warty, hoary, breezy, dapper,
edgy, feisty etc) and sends back the requested recipe(s) to the client
--- Client processes recipe file, creating a tar of all the needed info
---> Client connects, and sends the resulting file to the server (along
with details regarding which release and which recipes are in use if
stateless)
--- Server untars file, and processes it for viewing over the internet
site.

Obviously this could use a bit of refinement (for example, having it add
to a current set of data if another recipe is requested after, or
creating a log of changes between multiple commits of the same recipe)

I'd like comments etc on this, and whether you think it is viable and
useful to be something we can have in ubuntu.

Regards,
Martin Meredith
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part
Url : https://lists.ubuntu.com/archives/ubuntu-devel/attachments/20061102/396deca5/attachment.pgp 


More information about the ubuntu-devel mailing list