Unfortunate.

I don't think the UD client has enough diagnostic information to ever determine exactly what happened. Maybe one of the techs can be more precise.

Anyway, I can't see any fair way of awarding points for incomplete work units. After all, it's no use to the scientists.

Some encouragement though: this situation is very rare, and it won't impact your score in the long run.

Alther...

OK, I just got your post and I'll work on what you wanted to know. I do not want to answer right now as I want to take the time to check on the data. I think I need to be exact and precise as to what I shall give you later...

Stay tuned...

Alther...

I just tried to send you some data regarding your request.

However, I got the following message after clicking the SEND button:

[ServletException in:/contact_us.jsp] Exception thrown by getter for property problemTypeList of bean org.apache.struts.taglib.html.BEAN'

Please inform me if you got the message.

Didactylos...




You are taking a very mechanical view of things here and more importantly is missing a key point.

First: there was actual output.
It was not as if I wanted points in return for nothing. There was actual work done and even if that is valued to a naught, there was actual output as a result of that work. Only, that result was lost, and it was lost to none of my fault; it was lost because the UDMonitor software or something else screwed up. Up to a point, things were going fine; it was not as if things went wrong and I was not able to produce any output and now I am trying to get points from that. No sir, none of those. And it was not as if I had garbage results and try to get points out of it nonetheless. No sir again, none of those.

Second:
You and I know we are not working for dollars here, but ultimately there is something we want in return. For me, it is the points, for others - a pat in the back, or perhaps some cheering every now and then. All along we do it for a noble cause: find cure for diseases. Now, to try and subvert all those 'intangibles' into what I see as your tendency to view the world in a very mechanical no-money-no-honey if-you-don't-have-the-meat-then-get-out-of-here kind of arrangement, strikes me as out-of-line with what I see as the nature of what drives participants in this grid to contribute computing power--the argument for volunteerism.

As to the analogy that I used, I did not think you would read meaning literally. I used the money thing as a symbol to portray the impersonal, the mechanical, the material; I could have used the intangibles (like, if your girlfriend leaves you.., and those kinds of things) but it probably would not strike a chord with you; not with what I understand of your way of thinking.

andzgrid wrote:

I looked through the logs at the servers and could only find one entry for you which indicates something about what has happened on your machine. One of the agent files was altered by something other than the agent. The agent reacts by discarding everything and going to the server for all new files and fresh work unit. The file may have been damaged by a virus, bad disk sector, memory error, some device driver wrting to the wrong place in memory while the agent was also checking the file, or it could have been some problem with UD monitor and the way you are using it. So, at a minimum, you might want to run a full virus scan of your system just to be on the safe side. Also, a disk scan for surface errors might be useful.

Other crashes may or may not return anything to the server. In your case there were none that came back with anything. Crashes can happen from a large number of reasons, possibly due to any number of other software or hardware components on your system. However the most common are probably:

- Running out of total virtual memory because too much was allocated by all of the running applications on your machine, or that one or more of them had a memory leak. The cure is to run fewer applications or increase the maximum virtual memory limit.

- The DEP setting in XP SP2 or later, and certain firewall / antivirus products which attempt to prevent stack overflows can cause a crash.

- Running out of free disk space or losing permissions to write to the World Community Grid install directory can crash the agent.

- Deliberately ending one of the agent processes using task manager or similar tool

- Surprisingly often there might be a subtle hardware problem and sometimes only in the floating point portion of the processor, which is normally not used in common applicaitons. A thorough CPU test might find the problem. Increased CPU heat can make latent problems show up too.

- Memory errors can also cause unpredictable crashes. Try a thorough memory tester.

- Tampering with the agent and its data in some manner.

- Probably some other factors that I am forgetting right now.

There are faq entries about the above topics as well as quite a few forum threads.