SoC Proposal 2007

* Deliverables

At the end of the program, the code for the backend/frontend of the
job subsystem will be written to the extend it can handle:

- One job to one machine on a recurring and non-recurring way
- One job to all the machines on the network
- One job to all the machines sharing the same profile
- Multi-job error handling and notification to the user
- Job script pre-processing

* Project Details

Right now SCIRE's job subsystem on the client side can get jobs from
the server and run them in a simple way. All the features of this
project will be implemented on top of this so the client and the
client-server interaction could be kept as simple as possible.

The more complex tasks (as jobs for multiple machines) will be mapped
to a one-job one-machine and structured as a logical unit on a layer
approach. So we will have a two types of jobs, logical and "physical"
job; the "physical" jobs that make a "logical" job will be stored on a
different database table. This approach allows a flexible way of
dealing with jobs for multiple machines and how errors can be
treated. The success of a logical job will be set upon the success
of all the physical jobs attached to it.

For recurring jobs the entries on the database will be reused and for
logging and history purpose the job_history table will be used and
modified.

The UI will make simple the creation of jobs as it will involve the
following steps:

- Select a script

- If necessary: define some dynamic variables the script may
  require. The variables of the script will have a special syntax
  that will allow the system to preprocess them and create a form
  the user could fill them easily.

- Select a target

- Select, if needed, the recurrence

The existing code of the UI will be refactored to simplify the tasks.

The failed jobs will be reported on the main page so the user can
solve the problem and, if is a recurrent job, it will be paused
until the user fix it or it will continue.

* Profit for SCIRE

SCIRE will have a core component of its architecture almost completed
and will be capable of handle a wide range of common administrative
tasks. Other modules that need to perform an action on the client will
be possible given that they could use this subsystem to abstract the
interaction.

* Roadmap

from April 9 to May 28 - The current code-base for job management will
be studied and will be analyzed the way other systems deal with the
problem (e.g. the m23 online documentation isn't very clear about this
and seems to be very tied to package management.)

from May 28 to June 18 - any necessary modification to the client will
be performed, the database layout will also suffer modification so it
can support the logical and physical jobs. The recurring job
management coding will began.

from June 18 to July 14 - The recurring job management and the error
handling at backend level will be finished. All backend code will be
written. The UI work will began.

from July 14 to August 12 - The pre-processing script code will be
finished. The UI will also be completed and the whole subsystem will
be available throw the web.

from August 12 to August 20 - Test and bug-fixing

* Biography

I am a 21 years old student from Arequipa, Perú. Right now I'm working
towards my BA. degree on Ingenieria Informatica (Computer Science
equivalent) with expected graduation time at December 2007.

My first experience as a web application developer was a little
project for my university (http://inf.ucsp.edu.pe) which objectives
were make available throw the web all the research articles the
undergraduate students were into. This project allow me to learn about
mysql, php, smarty, html, css and also I used as a sand-box for
experimenting. My latest project is the web site for the Latin American
Center of Studies on Informatics (http://www.clei.cl).

I also was the administrator of a small lab on my university. My tasks
were migrated all the machines to Gentoo Linux and provide centralized
authentication and storage. Thanks to this experience I got in touch
on a deeper level with the system and also had to deal with job
distribution among all the machines. A little script did the job but
it was obvious that a more-secure, better thought solution was
needed. I am currently the administrator of the faculty server.

I feel a deep commitment with the free software and specially with the
Gentoo community, I work as a collaborator on the Gentoo documentation
Spanish translation Team and on the Gentoo Weekly Newsletter Spanish
translation team. I also worked as translator for the serendipity
weblog system. I see this as an opportunity to be more helpful to the
community and I would love to continue working with the SCIRE team
even after the Google's Summer of Code program concludes.

Thank you for considering this proposal.