I have a task that seems to be like the frog that covers half the distance to the edge of the pond each day! http://climateapps2.oerc.ox.ac.uk/cpdnboinc/result.php?resultid=20115492

Around 9pm last night BOINC told me it had six hours to go. Now over 12 hours later it still has almost three hours to go at 99.150% complete. Last time I saw this behaviour, the task seemed to get stuck after sending all its files and never actually completed. Will check back in a few hours time to see what is happening.

Based on the last hour, the remaining 2.5 hours indicated by BOINC will take about ten hours. I notice that the task folder is 1.6GB in size. Most of the time they only get that big when I have network activity suspended for other diagnosticor other reasons.

So average is currently 3.3 hours/1% completed. This last 1% has now taken about 20 hours and still a bit to go! Has anyone else seen this behaviour? Is it just part of BOINC not being very good at working out how long things should take when more than one project and various types of task are involved? Or is there something funny about this task?

Hi Dave

I think it may the many different types of tasks. There's only one Task duration correction factor per computer, which can get muddled when several tasks with very different durations are being run.
But this is getting into the area where I don't think too well these days.

Thanks Les, sadly I didn't have checkpoint debug enabled for the log till a few hours ago so I can't compare the rate earlier in the task with now which would have told me whether things had really slowed up or not. It is now down to about half an hour to go according to BOINC at a tad over 99.8%

Edit: I also note that at least twice in each minute (BOINC estimated time remaining) it drops by maybe 40 seconds and then adds 30 seconds back on.

Dave -

I notice the workunit you referenced earlier in this thread completed successfully. I also notice that most (if not all) of the workunits on this computer (1418333) get suspended A LOT!

Have you considered changing some BOINC preferences so that doesn't happen? Maybe the workunits would make more steady progress toward the endpoint if they weren't interrupted so much.

---

Have you considered changing some BOINC preferences so that doesn't happen? Maybe the workunits would make more steady progress toward the endpoint if they weren't interrupted so much.

I will have a look at the settings to see what is different with that box and the one I am typing on, both of which have a lot of suspend requests whereas another (currently dead till I replace PSU) which didn't seem to have any. - If I can't spot the problem easily, may have to wait till new psu on the dead laptop. All three computers in question are currently running boinc under WINE due to lack of native linux tasks.

Edit:Not seeing anything obvious, the computer in question is only running CPDN 99% of the time and has 2GB of memory/core compared to only 1 on the machine that didn't have all the suspend requests. I can understand it a bit more on my desktop which gets used on a daily basis.

Suspend when non-BOINC CPU usage is above ?

Suspend when non-BOINC CPU usage is above ?

Not ticked on any of my boxen. Not spotting anything obvious that might account for it and unless I am mistaken, when I get the new psu, will find the box that doesn't have lots of suspend requests has exactly the same settings.

Edit: The two running tasks on the computer in question that have uploaded at least one zip don't show any suspends. I wonder if it is only happening towards the end of computation?