I noticed a couple of WUs that had a second task sent out presumably because the first task timed out. The first task however was returned and arrived before the second task was returned. Consequently the second task was marked invalid merely because a valid result was already returned.

example: https://www.mlcathome.org/mlcathome/workunit.php?wuid=5465870

It is a bit frustrating but if after the second computer is allocated the job, the first time the first computer reports back to MLC is with a completed task, it would be silly to throw it away.

Its not helped by users stacking up loads of jobs, in this case he has 360 jobs stacked up on an average turnaround of 7 days with 8 cores and regularly has jobs timing out (roughly 40%).

@pianoman, do you not think this is an issue ?

Time and effort are being wasted on returning what look like perfectly good results only to have them marked as invalid because the original work unit gets returned shortly before a replacement work unit can get returned. By all means cancel duplicates if they haven't started but if it's running and then completes as a valid task it should be given credit please. I do not have a problem with the original task getting zero credit if it is returned after its deadline and already completed by another.

If the project doesn't think the time and effort is worth credit for these results then I'll go along with that and just abort all non-original tasks that get downloaded to ensure I'm not going to run into the problem.

Thanks