Hi Guys!
I run your tasks in corporate network at 20+ PCs.
BOINC version is 6.2.14 (latest stable).
Besides Malaria, I'm conncted to Rosetta, WCG, ClimatePrediction.
Please consider this report as official complain on poor quality of tasks you distribute across grid members, as in last 30 days half of tasks to be processed were fatal errors (GPFs) or wrong source data (where task status is 100% complete, but BOINC still thinks is should continue calculations and the only way to stop it is to manually cancel the task execution).
Separate story - is extremely poor source data for "estimation of parameters of infection dynamics tasks", where more than half of tasks ends with "fatal errors" like division by zero or resulted in "wrong calculation".
Your project is the only one among others who totally hangs user PC with bad resulting status and bad source data submitted for client processing.
PLEASE DO SOMRTHING TO STOP THIS BULLSHIT, IF YOU REALLY GONNA TO ACHIEVE SOMETHING IN YOUR RESEARCH,
OTHERWISE YOUR PROJECT WILL BE DISCONNECTED FROM ALL OUR MACHINES AND I WILL REPORT TO COMMUNITY WITH SCREENSHOTS AND MORE DETAILS HOW RESPOBSIBLESS YOU ARE WHILE PREPARING AND DISTRIBUTING COMPUTATIONAL TASKS ENDING IN NOTHING EXCEPT RESETS AND RELOADS....
This really annoys a lot of users and makes participation in your project useless and targetless...
Source data for remote processing should be bug-free and very valid, like WCG or Rosetta does.
Good luck!
/ 
Hi Guys!
I run your tasks in corporate network at 20+ PCs.
BOINC version is 6.2.14 (latest stable).
Besides Malaria, I'm conncted to Rosetta, WCG, ClimatePrediction.
Please consider this report as official complain on poor quality of tasks you distribute across grid members, as in last 30 days half of tasks to be processed were fatal errors (GPFs) or wrong source data (where task status is 100% complete, but BOINC still thinks is should continue calculations and the only way to stop it is to manually cancel the task execution).
This is an issues that cannot be changed but has been written extensively about. Try here https://malariacontrol.net/forum_thread.php?id=781
The "estimate" is just that a pure guess and is ALWAYS wrong. The way the OPT units work is that they "estimate" how long they will take and how far along they are but can't really tell because they are not written in the standard Boinc way. Therefore they think they are at 100% but in fact will continue crunching for some time. Each unit will take about 2 hours of actual crunching time.
{quote]Separate story - is extremely poor source data for "estimation of parameters of infection dynamics tasks", where more than half of tasks ends with "fatal errors" like division by zero or resulted in "wrong calculation".{/quote]
You can turn off the OPT units in your preferences.
{quote]Source data for remote processing should be bug-free and very valid, like WCG or Rosetta does. Good luck! [/quote]
Each Project is run independently of every other Project, otherwise you could go to one website and see all other projects forums. That means that some are "good" and some are "bad", which of course is all in the eyes of the beholder. I think this is a "good" website, the Admins are receptive to constructive criticism, they are actually here and do not leave everything to others, and actually give feedback on a regular basis.
____________
/
Hi Guys!
hi
I run your tasks in corporate network at 20+ PCs.
BOINC version is 6.2.14 (latest stable).
That's great, thank you for the effort, and sorry for causing you so much trouble. It's true that our error rate has risen in the past, but on average it is far away from the dimensions that you reported ("half of all crashing"). Following your post we looked at the results you sent back, and realized that there were some of those errors described here:
click
going through the results database revealed that half of our errors (I'm only talking about the optimizer application) seems to be exactly the one described in the thread above. Apparently the clients from version 6 are a new generation, something with user rights is different, and your client 6.2.14 has a bug (see link above), which is apparently fixed in 6.2.18..
It looks like the main malariacontrol app is not affected much by it (still investigating). But our "optimizer" app is more sensitive, since it uses java and all that.. we're somewhat pioneering in the use of java with boinc, and have to deal with a few "child diseases", it seems..
nevertheless, the errors you describe are probably most due to that bug in the core client (correct me if I'm wrong)..
as a countermeasure we restricted sending work of the optimizer app to core clients before version 6.0.. for the moment.. and want to see what effect that has.
Your project is the only one among others who totally hangs user PC with bad resulting status and bad source data submitted for client processing.
we would like to know more about that.. if your computer hangs, that's extremely difficult to notice for us, since we're probably not getting anything back, and it just appears as a "no response" case. Please check if it happens again (..would mean it has nothing to do with the error above, but is a separate one).. if you see it, it would be very helpful to have the workunit name.
apologies again, for inconveniences. If you still have problems with the "optimizer" app, consider opting out of crunching it, using the settings in your userprofile. I think the main malariacontrol application is more stable)
regards
Michael
____________
Michael
/