Work unit stuck at 100%

[wertperch]

I have a work unit that has been stuck at 100% for at least a couple of hours. It did the same yesterday, and I restarted the client, only to discover that it was (apparently) only at about 9%.

The original estimate for completion was just under 10 hours - it's up to 9:57 now, stuck at 100% but still counting (and still taking its alloted CPU time!)

So my question is this: do I assume that this work unit is somehow bugged or broken and abort it, or leave it alone?

For reference, it's cms_GA[e20-30MB_Lin64s]_Oryza-sativa-Japonica-Group_CM000145.lin.EMBL_RF00824_mir-50_1284579226_49941_0 and it's running under Vista on an Intel CPU T5300 @ 1.73GHz

[yoyo]

If the workunit still consumes CPU at 100% (which is mostly the case), let it run.
Don't restart a workunit. They do not have checkpoints, so the start from the very beginning.
yoyo

[wertperch]

Thank you for this. I will leave it running, but am not happy that the application seems to me to be flawed. I have read opinions from others that larger work units seem to have more problems, which does not fill me with confidence!

This is the offending work unit, by the way:

If the workunit still consumes CPU at 100% (which is mostly the case), let it run.
Don't restart a workunit. They do not have checkpoints, so the start from the very beginning.
yoyo

[Michael H.W. Weber]

Well, even the reported problems for a very small subset of (very long running) WUs has been resolved with the new binaries we released for Windows 32/64 today.
The apparent "stuck at 100%" is no bug but for some projects has at least in part to do with BOINC's run time management and the fact that run time prediction with some of the RNA World WUs is difficult since the code involves some randomizing elements. As yoyo said, in case of doubt, just make sure the task still consumes CPU time.

Michael.

[wertperch]

Oh, thank you!

Well, even the reported problems for a very small subset of (very long running) WUs has been resolved with the new binaries we released for Windows 32/64 today.
The apparent "stuck at 100%" is no bug but for some projects has at least in part to do with BOINC's run time management and the fact that run time prediction with some of the RNA World WUs is difficult since the code involves some randomizing elements. As yoyo said, in case of doubt, just make sure the task still consumes CPU time.

Michael.