LHC@home

I was just wondering...

I'm on holiday for a week and have aborted two tasks (before starting) since the deadline would have expired before my return. I know these get pushed back in the queue for reassignment but I was wondering just how quickly this happens.

My understanding is that when a Task is initially assigned to a Client it goes to two Clients at about the same time. If the aborted Tasks are put on the bottom of the stack then the other Client will have a unverified Task hanging around for a long time.

Do Aborted Tasks get a reassignment quicker than a Task that fails with Error?

Thanks.

Judging by the state of my pending list, they - currently - go to the end of the queue and stay there. WU 2704882 has a replacement task created on 23 August, still unsent.

We had a discussion about this a while ago. BOINC does have some mechanisms for speeding up the resends, but at the time the side effects were worse than the disease - reduced deadlines, and not every host being (at the time) deemed 'reliable' because of a spate of bad workunits.

At the moment, the "back of the queue" is also further away than normal, because the administrators released a large batch of work all in one go, to buy themselves some thinking and programming time without having to tend the server every day. We're getting through them quite well, so we should reach the resends at the back before too much longer.

Thanks for the quick reply.

I've been crunching for SETI for many years and as I recall they had a similar issues at one time. It sticks in my memory that they managed to resolve it so that Aborted Tasks and/or Error result Tasks got pushed into the queue someplace down from the top but at a spot in the queue so that validation of a completed Task didn't take forever. I think it was an issue of necessity with them because they have always had long queues in there server waiting to be sent to clients.

Thanks again.

Cheers,
Tom