... This morning we generated another two tasks with short deadline (2d) for each of the remaining "Workunits without canonical result" (13k at that time) to quickly finish these.
At the moment (03:50 UTC on 29th May) there are still 3196 tasks in progress but only 7 workunits without a canonical result. At the highly unlikely very worst, each of those 7 could have 4 associated tasks still 'in progress'. It would be more likely that each has 3 tasks and it can't be less than three, otherwise an additional task would have been resent by the scheduler to keep up the initial replication of 4 or it wouldn't be still in play 'without a canonical result'.
I know this is stating the obvious but I wanted to point out that, of the 3196 tasks still in progress, only 7-14 will make a difference and those will come from probably just 21-28 tasks. In the interests of not wasting computing resources, I would suggest to volunteers in general that if they have tasks still in progress, click on the workunit ID on the website and see if there is already a canonical result. If so, you could abort such tasks, particularly if such a task hasn't started. Over the last day or two, I've been doing this and have aborted more than 200 such tasks over my hosts. Of course, if you want the credit for a partly crunched task, you would allow it to finish.
The main reason for posting was to check my reasoning about when FGRP4 could be restarted with the new app. Am I correct in assuming that as soon as there are zero workunits without a canonical result, irrespective of how many 'in progress' tasks still remain, the flow of new work could be restarted? After that, the 'old version stragglers' could time out or be returned 'at their leisure' so to speak. If so, is it likely that the flow will be restarted later today, or is it going to be after the weekend?
The transition is done, new app versions released, new work generated.
BM
Is it intentional that the new work is being issued with a rather short deadline?
Replication appears to be back to 2 (good), but I'm getting "high priority" warnings as the work appears to be issuing with deadlines 3 days from issue. If not needed, this causes some extra confusion, disruption, and concern among the participants.
EDIT: I hadn't noticed Peter's report whilst I was composing a rather long winge about short deadlines and BOINC's propensity (some versions anyway) to stuff things up. In the end I cooled down enough to delete most of the winge and when I posted what was left, I saw Peter had beaten me to it.
RE: ... This morning we
)
At the moment (03:50 UTC on 29th May) there are still 3196 tasks in progress but only 7 workunits without a canonical result. At the highly unlikely very worst, each of those 7 could have 4 associated tasks still 'in progress'. It would be more likely that each has 3 tasks and it can't be less than three, otherwise an additional task would have been resent by the scheduler to keep up the initial replication of 4 or it wouldn't be still in play 'without a canonical result'.
I know this is stating the obvious but I wanted to point out that, of the 3196 tasks still in progress, only 7-14 will make a difference and those will come from probably just 21-28 tasks. In the interests of not wasting computing resources, I would suggest to volunteers in general that if they have tasks still in progress, click on the workunit ID on the website and see if there is already a canonical result. If so, you could abort such tasks, particularly if such a task hasn't started. Over the last day or two, I've been doing this and have aborted more than 200 such tasks over my hosts. Of course, if you want the credit for a partly crunched task, you would allow it to finish.
The main reason for posting was to check my reasoning about when FGRP4 could be restarted with the new app. Am I correct in assuming that as soon as there are zero workunits without a canonical result, irrespective of how many 'in progress' tasks still remain, the flow of new work could be restarted? After that, the 'old version stragglers' could time out or be returned 'at their leisure' so to speak. If so, is it likely that the flow will be restarted later today, or is it going to be after the weekend?
Thanks for any info.
Cheers,
Gary.
The transition is done, new
)
The transition is done, new app versions released, new work generated.
BM
BM
RE: The transition is done,
)
Is it intentional that the new work is being issued with a rather short deadline?
Replication appears to be back to 2 (good), but I'm getting "high priority" warnings as the work appears to be issuing with deadlines 3 days from issue. If not needed, this causes some extra confusion, disruption, and concern among the participants.
RE: Is it intentional that
)
Nope. Sorry, fixed (for newly generated tasks). Thanks for reporting.
BM
BM
RE: .... new work
)
Do you realise it only has a 4 day deadline?
EDIT: I hadn't noticed Peter's report whilst I was composing a rather long winge about short deadlines and BOINC's propensity (some versions anyway) to stuff things up. In the end I cooled down enough to delete most of the winge and when I posted what was left, I saw Peter had beaten me to it.
Cheers,
Gary.