I suspended some running WU (I run SETI and Rosetta) while I was working on PC, which involved having to reboot.

When BOINC restarted, the SETI unit picked up where it left off, but the Rosetta unit restarted at 23% rather than the 96% it was at previously.

Why?

Any time you reboot, some work is lost. BOINC applications preserve their work through a process called checkpointing. Some applications (perhaps SETI?) checkpoint frequently. Some cannot. In Rosetta's case, it depends upon the methods being used in the task you have. Some methods checkpoint frequently, and others cannot. Over time, more checkpointing is being added to all methods.

So it sounds like you probably have a relatively short runtime preferences (and so the 73% of a task you lost is still only an hour or two) and that you were running a task that cannot checkpoint frequently.

There are two related controls you have. One is the computing preference to leave suspended applications in memory. If you tell BOINC to snooze, or it preempts Rosetta to run SETI for a while, this preserves work by retaining the virtual memory the task was using.

The other is a setting for the most you will allow a BOINC application to write to disk. It defaults to 60 seconds. Meaning that even if the application wants to, BOINC will not actually use the disk drive any more then every 60 seconds to write the checkpoints. Unfortunately, this does not cause a write to occur every 60 seconds. It is just a way for you to prevent a BOINC application from using too much of your disk time.

---

Rosetta Moderator: Mod.Sense