This business of long times between checkpoints seems to be giving many crunchers a headache. Using hibernation rather than a normal shutdown saves the present status of the agent and allows it to resume from where it left off at next power-up without losing any work.

In another thread I described how I set up hibernation and power-up to occur automatically on some of my machines. I think many users have been frightened by the complexity of setting their PC's up in this manner. However, you only need to make the process automatic if you need unattended hibernation and/or power-up sequences. For users who normally power down their machines themselves and then boot them up again manually it is really very simple to implement a hibernation strategy. The following comments assume Windows XP-Pro.

At present when they wish to shut down the PC they will click the START button, select "Turn off Computer" and then "Turn off". The procedure for going into hibernation mode is just as simple: Click the START buton, select "Turn off Computer" but then click the "Hibernate" button instead of the "Turn off" button. The PC then saves an image of its present status to the hard disk before shutting itself down. When powering up the next time it automatically loads the disk image back into memory and continues processing from where it left off. On some PC's you have to press the SHIFT key to show the Hibernation button. Without the SHIFT key pressed the "Hibernate" button may be replaced by the "Stand-by" button.

If the "Hibernate" button is not present in the shut-down menu then it needs to be enabled (this is a once-only requirement). Go to Control Panel, open Power Options and select the Hibernation tab. Make sure the "Enable Hibernation" checkbox is ticked, and click "OK".

David.

Thanks Lawrence,

I agree with everything you posted.

David.

You seem mixed up. Hibernation is not the same as standby.

In hibernation, the computer saves all the contents of memory to disk, then shuts down completely.

The techs have determined that it will take too long to rewrite the code with more checkpoints just for these problem work units. They ask for your patience until we get them finished.

Hello padre999,
Sekerob has a post in Start Here that talks about the way check points are handled by the various science applications: http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=11332
Last year I suggested that the time of the most recent check point be displayed on the graphics.

Lawrence

The above config file illustrates that multiple things can be entered, and the information combined. I may do some of that myself, but entered the absolute minimum for checkpointing. Go to the BOINC manager help to find out more.

Even then, nothing happened. After exiting BOINC and restarting BOINC, thus ignoring another checkpoint checkpointing begin to appear in the messages.

It also helps to notice the different of the first one from the beginning, or to see at least a couple of them to know from experience how long you have to wait for the next one on the same project on the same computer. Different workunit batches can still cause variations. They are about 10 minutes apart on this machine. CALL A DOCTOR! BOINC is having a baby!