BOINC

BOINC is an example of this approach. There are a smallish number of projects being processed around the world right now using this approach. I have joined one called the Search for Extra Terrestrial Intelligence. Massive quantities of data are collected from the sky and sliced up into small chunks of data. These chunks are sent to hundreds of thousands of peoples' computers around the world. There they are processed and returned to the SETI team. Who knows one of might just yet find that signal from outer space that proves there is life over and above ours here on earth.

SETI is one example of about 20-30 projects that I know of that uses BOINC. Personally I have joined ten projects but do not process them all at the same time; you can share your processing power out around your favourite projects. You can join in with any of these projects if you wish. Here is the link to the many of the projects underway or just getting underway. Note there are some serious people "processors" and teams out there who take this very seriously. Some teams have hundreds and thousands of PCs all piling in results....its quite competitive but equally its also nice just to make a contribution.

SETI of course is the search for extra terrestrial life.

I have also joined the CPDN project. This project is modelling climate changes and is run by Oxford University.

Finally I have joined the LHC project. This is a CERN project and the work is modelling the Hadron Collider which is under construction in Switzerland. This project will, from 2007, fire protons around a ring that is 22Km long and 90 Metres underground. They will bring about the collision of these protons by adjusting their course through the use of thousands of powerful magnets. This will re-create the first few seconds after the Big Bang which, it is believed, created all we know today.

In joining these projects I hope to:

• Help find other life that will share their technologies with us to help mankind survive.
• Help to predict weather condition over the next 100 years so we know how long we have before we really must find life out there to save us and our planet
• Help explore what new energy sources can be available to us as a bi-product of colliding protons and all that it leads to so that when we find life out there we can travel to share science. At the same time, we can use new energy sources to stop killing our planet in the way we do.

1). I established that there were peering problems between Cogent and (inter alia) France Telecom. A successful thread to help thousands of seti users was established along with a web page to guide people through the use of proxy servers to overcome the issue. I was helped by other seti users who are acknowledged on the web page giving help. Some 14000 accesses were made to the thread and over 20000 accesses to the web page. The work was acknowledged by the seti project in their news pages.

2). I did a code inspection to try and understand why the seti client produced some 6.5M OS page faults for a typical work unit. This is unnecessary stress for an OS. I determined that the use of dynamic memory allocation was inappropriate and made a trial modification which reduced this activity by three orders of magnitude. I passed this work to the seti team; the new seti clients now use dynamic memory as I had suggested.

3). I observed that the cpdn model I was processing failed after a ntp resynch. Random fails had been discussed across the threads of all BOINC projects but no one could discover why they happening. My observation was reported and various people tested my theory that the system clock, when adjusted on a routine resynch, caused the running model to fail because its understanding of time steps changed outside of the model. All testers were able to reproduce my findings and the BOINC project was alerted. A change has since been done to make model time independent of system time. This will prevent, I think, a very large proportion of model failures in the future.

4). A number of users were having trouble connecting in early and mid August. There were three things going on that looked interesting. People could work through proxies but not direct. Some people could ping and trace route but others could not. The seti Servers very feeling unwell! After some thought and investigation with other seti users : two ISPs had faulty DNS entries after UCB had done a small dns change the previous week. Cogent had routing table problems which prevented in bound traffic. UCB had a  mis-config themselves with a public facing server. All were corrected and guess what.....it all worked OK again!

5). I might have helped Matt Lebofsky (UCB) with the server error 500 problem that has plagued many users. After many email exchanges between us talking about various parameters he decided to upgrade apache and fastcgi to the latest versions and hey ho all error 500s ceased! Well he and I didn't talk about an upgrade (we talked everything but tbh) but anyhow I feel I at least brought the matter up and the whole thing ended in success. Anyhow most folk feel OK with the result and the server now does not fail 5% of all updates from clients! So that's a bit less work for them!

6). I recently started running boinc server - 5.1.3 under Linux. I upped the reporting level in php.ini and detected that cookies were not being written properly because header output had already commenced due to a subscript error message being sent. Well actually cookies are probably OK without error messages being sent but there is clearly a subscript error in translation.inc. Passed to boinc_dev and cause is found and fixed! Hey all in a couple of hours too.

7). I am currently trying to build a set of Boinc users that can provide small proxies for the community when network problems hit the project. All help welcome. See my proxy stuff at the bottom of this page.

8). Suggesting boinc management port should be agile. Accepted by Dr A and mod released. 1043 no longer sacred!

9). Suggesting support folks should have server side scripts to examine scheduler logs to help diagnose problems for users. Done by Bruce Allen - thanks Bruce.

v 4.32 Std  Version

v 4.32 Tigher Version