Hey, what is up with this WU?

Simply put, the task result output from you host was not within the error tolerance band once the quorum formed (two tasks strongly similar) and the canonical task was selected.

Keep in mind in cases like this where the WU goes to more than the initial replication, all the other eligible tasks which remain after the quorum forms are compared to the canonical task and must be at least weakly similar to it to have credit granted.

Fortunately, a case like you saw here is very rare on EAH, and I'm not quite sure what the story was to bring it about. This WU dates back to just before the shutdown to upgrade the backend for R4. So if I had to guess, I'd say that the output file for your host got 'lost'. This is why it was sent to extra replications in the first place, since there was nothing for the validator to compare with when the wingman reported.

Alinator

No, 'Zombies' don't seem to be a problem with EAH, like they can be with SAH from time to time. Also keep in mind that updates to the pending list can be out of sync with the other summary pages by a bit. Some times a page refresh gets things moving for it.

AFAICT, this one is still in play since the scheduler has it has already 221'ed the two extra replications it didn't need.

EAH is just less aggressive about purging the BOINC database as quickly as some other projects are. My guess is the project will poof this WU from the record in a few days.

On second thought after looking this one over again, it might be a zombie. It was definitely having some problems back during the transition over to R4. So if it doesn't go poof in another week, then it might be worthwhile bringing it to Bernd's attention.

Alinator