It sounds like normal variance to me.

One way to get some more detailed information would be to check your device statistics page: http://www.worldcommunitygrid.org/help/viewTopic.do?shortName=points#24 (if you click on the device name, you can check the history over the last few weeks). Hopefully that will let you spot any problem devices.

If you pin down a problem, please post again if you need further help.

Firstly, we are seeing some larger than usual work units currently. Normally, you get full credit for the extra time on a long work unit. However, if crunching is interrupted, and it has to restart at a checkpoint, you don't get credit for the lost time. Large work units often have much larger gaps than normal between checkpoints, so if you shut down your computers, you may lose quite a lot of time.

If your work units were actually failing in some way, you would see the opposite - you would see lots of work units being completed unusually quickly. I don't think that's happening here.

I would suggest switching these troublesome devices to run only one of the projects with short work units, using a custom device profile. Genome Comparison has reliably small units. Have a look at this graph to get a sense of work unit sizes: http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=13733

Sorry, you can't do anything about the extra device registrations. You get a new one every time you reinstall the agent.

Hello people. It is me again, reporting.

I have been able to trace the problem on these computers to the fact that the computer is continuosly rebooting, thus losing quite a bit of the work done. I know by know that the problem seems to be that UD interacts quite badly with another application running in those computers. The other application is a monitoring tools, that receives updates constantly via UDP and makes request to a map server. When both the application and UD are running, the machine will reboot shortly after activation of the screensaver. If I deactivate the screensaver, then the machine will reboot randomly every 45-120 minutes. The computer runs smoothly when either only the ud or the application is ran.

As we have recently upgraded the application, I tought it may be some contention or cpu sharing problem, but it does not seem to be the case, as the last reports from snmp show consistently that the process using more cpu (and according to my logic, the use of cpu in that particular recolection cycle) is the UD process; as well, there are no pagination issues, as virtual memory seems to be just very rarely used.

UD seems to run fine using a previous version of the monitoring application. However, I can not keep using that version as it uses TCP instead, and a new and strict policy on the network management prevent me from using too much tcp conections.

The funny thing is that UD and the very same application seem to run finely on other computers using same processor and motherboard, and only difference being those others computers having less memory (and, no, it is not a memory problem, as I have already hot tested it for more than 72 hours.)

I am quite lost, and do not know where to walk now. Any suggestions?

Best regards,
Chilango Chilaquil

Hey Ya Sekerob!
Had a total system failure. Had to reload everything from scratch. How do I download a new agent? I want to stay with UD because I have an older system.

It happened to me ... random reboots ... power supply replaced last year - no problems since.