Status update

Message boards : Malaria Control : Status update

Author Message
Profile maire
Volunteer moderator
Project administrator
Project developer
Project scientist
Send message
Joined: Nov 7 05
Posts: 437
Credit: 118,125
RAC: 0

We have downloaded the first set of predictions to our master database today. Apart from a scary peak in the server load this went fine.
We are still working on the next client version, which will include some of the developments in the science application. We expect to update the client to 527 next week.
Nick
____________
Nicolas Maire
Swiss Tropical and Public Health Institute
http://www.swisstph.ch

AnRM
Send message
Joined: Mar 7 06
Posts: 54
Credit: 2,130,571
RAC: 37

Thanks for the update, Nick....good stuff!!....Cheers, Rog.
____________

Profile Bill
Send message
Joined: Jun 21 06
Posts: 11
Credit: 452,831
RAC: 124

Nicolas

Can you give us an update on Client progress and development? What is happening in Beta these days?

Bill Freauff
____________

Profile maire
Volunteer moderator
Project administrator
Project developer
Project scientist
Send message
Joined: Nov 7 05
Posts: 437
Credit: 118,125
RAC: 0

This week was mainly dedicated to work on the server code. Also a few scripts for managing the master result database. On the science application side, the only development is a monitoring thread that kills a workunit if writing a checkpoint takes too long. This should solve the issue of stuck workunits (very rare indeed it seems, but still very annoying).
Nick
____________
Nicolas Maire
Swiss Tropical and Public Health Institute
http://www.swisstph.ch

Chris Sutton
Send message
Joined: Nov 10 05
Posts: 296
Credit: 4,941,683
RAC: 0

Thanks for the update Nick.

This should solve the issue of stuck workunits (very rare indeed it seems, but still very annoying).

I used to get this fairly regularly on my linux box (about 4 times a week). Since the update to v5.36 it has not happened once. I suspect something in the additional debug code is preventing the problem from happening now. Perhaps slowing down execution of the app by a few milliseconds has helped?

Gimlei
Send message
Joined: Jun 20 06
Posts: 4
Credit: 9,687
RAC: 0

Thanks for the update! We love receiving them, they keep us interested and motivate to give more crunching time to MC.net :)

By the way, are there any estimates, on when could MC.net go "production stage"? Or rather, which steps are still remaining to be taken on this way? What is still required to be done, both scientific- and client/server-wise?
____________

Profile maire
Volunteer moderator
Project administrator
Project developer
Project scientist
Send message
Joined: Nov 7 05
Posts: 437
Credit: 118,125
RAC: 0

I used to get this fairly regularly on my linux box (about 4 times a week). Since the update to v5.36 it has not happened once. I suspect something in the additional debug code is preventing the problem from happening now. Perhaps slowing down execution of the app by a few milliseconds has helped?


Interesting idea! I did not know that it seems to happen less often with 536. For v5.36 I used an updated version of the Fortran Compiler on linux, maybe this also helped some. I hope that the workaround in the next app version will do the job (kill stuck wus). We would then also have a way of knowing how often this happens, from the stderr output. The new app version is under way, but will not be ready before next week.
____________
Nicolas Maire
Swiss Tropical and Public Health Institute
http://www.swisstph.ch

Profile maire
Volunteer moderator
Project administrator
Project developer
Project scientist
Send message
Joined: Nov 7 05
Posts: 437
Credit: 118,125
RAC: 0

We have now got back all results for runs 4 and 5, and are currently analyzing these. Run 7 is still missing a few results.

Here are some of the issues that need to be addressed before we can open for new users or go into full-scale production:
1. There still some bugs we are sorting out.
2. We are recruiting a new team member but he has not started yet. When he does this will help us progress faster.
3. Before we open again for a substantial number new users, we will new to upgrade the server hardware.
____________
Nicolas Maire
Swiss Tropical and Public Health Institute
http://www.swisstph.ch

Profile Bill
Send message
Joined: Jun 21 06
Posts: 11
Credit: 452,831
RAC: 124

Mairie,

Thanks for the updates !!

Do you think that any of the Client version changes will reduce the number of work units that fail with Client Error status ? Or the issue where the Client is not correctly shut down during a Windows Shutdown or Restart ?

Your updates are wonderful and we look forward to them !!!

Bill Freauff
____________

Murray
Send message
Joined: Jun 20 06
Posts: 3
Credit: 8,184
RAC: 0

Okay...whats news, a month down the line?

Regards
Murray
____________

Profile The Gas Giant
Avatar
Send message
Joined: Mar 7 06
Posts: 1207
Credit: 3,163,590
RAC: 2,556

An update from the recent BOINC get together.

AnRM
Send message
Joined: Mar 7 06
Posts: 54
Credit: 2,130,571
RAC: 37

Thanks for the 'heads up'.....interesting reading and a nice presentation from Nick. It's also nice to put some faces to names on various projects, BOINC etc.
____________

Profile maire
Volunteer moderator
Project administrator
Project developer
Project scientist
Send message
Joined: Nov 7 05
Posts: 437
Credit: 118,125
RAC: 0

I'm back from the BOINC workshop in Geneva, where it was great for me to link up with the guys from BOINC and the other projects. Thanks for linking to the proceedings and the mcdn presentation!
Next week we'll proceed with run 11. We now have a lot of data back for run 9 that should be interesting to analyze.
Thanks
Nick


____________
Nicolas Maire
Swiss Tropical and Public Health Institute
http://www.swisstph.ch

Post to thread

Message boards : Malaria Control : Status update


Return to malariacontrol.net main page


Copyright © 2012 africa@home