Return of the long running multi cpu w/u


Advanced search

Message boards : Number crunching : Return of the long running multi cpu w/u

Author Message
nairb
Send message
Joined: 21 Sep 15
Posts: 11
Credit: 47,810
RAC: 0
    
Message 6082 - Posted: 26 Jan 2017, 16:49:33 UTC

After a good run of multi cpu w/u I seem to have one that uses all the memory and cpu usage is almost zero. Its going to take ages to finish. It creeping along.

The question is... is this a valid w/u. Is it worth letting it run to completion??. Lots of disk activity as well.

Whats to be done??.

Ta
Nairb

PHILIPPE
Send message
Joined: 24 Jul 16
Posts: 84
Credit: 53,413
RAC: 0
    
Message 6083 - Posted: 26 Jan 2017, 18:16:41 UTC - in response to Message 6082.

Hi , Nairb ,

before taking a decision : there are 3 points to look after :1° cpu , 2° difference time elapsed and time run , 3° login screen of the vm.
For you :
-No activity cpu is no good.(1° point)
-Can you say if when you click on property on the left panel of boinc , the time of the last saving point , the time cpu and the elapsed time ?(is there a big difference between them)(big difference is not good)
-Can you open the VM , allways on the left panel of boinc client to see what appears in it?
If it 's this

it could be good


Disk activity can come from other processes or because ram is not enough.

After all this checkings , you can more precisely choose the future of this wu.
But it's always you who decide in last .

PHILIPPE
Send message
Joined: 24 Jul 16
Posts: 84
Credit: 53,413
RAC: 0
    
Message 6084 - Posted: 26 Jan 2017, 20:16:05 UTC - in response to Message 6083.

Finaly, you decided to wait and you had done the good choice:

Even if it was apparently a faulty wu ,because run time (85,233.83 sec)is very different from the cpu time (19,031.91 sec), the long wu ended and had been validated .
So you earn 693 cobblestones.

It's always difficult to advise someone because each wu is different but sometimes there is no doubt that a wu won't finish (for instance when the save point is not recorded in the wu's propriety).

nairb
Send message
Joined: 21 Sep 15
Posts: 11
Credit: 47,810
RAC: 0
    
Message 6085 - Posted: 27 Jan 2017, 3:01:18 UTC

Thanks PHILIPPE, I clicked on property on the left panel of boinc and saw that there had been only 2.5 hrs of cpu time in the last 20 hrs and it expected to last another 10+ hrs.
So I decided to close boinc and wait until all disk activity had stopped and restarted Boinc.
And after a little while the cpu activity rose to 100% and the task completed after another hour or so.

I guess the restart cleared something. At least it validated ok so something as gained. I have another multi cpu now and will watch it closely. We shall see.

Thanks for your help
Nairb

Message boards : Number crunching : Return of the long running multi cpu w/u