search

jtouug / freehal

Automatically exported from code.google.com/p/freehal
0 stars 0 forks source link

NCI / CI and switching around creates many problems. #1

Open GoogleCodeExporter opened 9 years ago

GoogleCodeExporter commented 9 years ago 
I hope it is the right place to complain about this issue.

There are several problems with NCI / CI and switching around.
If you have made your selection and want to change it, it does not work as 
intended.
Let's say you want to change from NCI to CI. When you do that, it seems that 
you will get CI workunits but they are running as NCI (status is running non 
cpu intensive). To fix this you'll have to detach project and add it again. 
Nothing else resolves this.

There is at least one more problem, and this is a big one!
I have a setup to receive 1 WU per core and CI. Now, everything seems to run 
fine for some time. But then for now apparent reason CI changes to NCI (I did 
not do that!). Now, the problem is, that when this happens and you have queued 
let's say 500 WU-s (like I had), then they ALL start to run simultaneously. 
Yes, all 500 WU-s will run AT ONCE.
What that means is your computer will totally hang! It comsumes all CPU, and 
all memory. Then OS starts heavy swapping which saturates all HDD I/O. So your 
computer will not respond to anything, all that you can do is pull the plug. 
And after that you'll have to come up with some clever plan to stop all those 
WU-s from starting again when you'll restart (I have selected to run always).

Now, after you have come up with a brilliant idea to stop all those WU-s from 
starting, you still have 500+ WU-s in your queue. If you do not want to abort 
them all, you'll have to suspend them all and start resuming them in small 
quantities to not overload cpu and memory again. That's quite a lot manual 
monitoring and resuming. Might take several days.

All this happened to all of my computers just some days ago. And this is not 
the first time!
I home you can do something about it, as it is very, very annoying when it 
happens.
All I want to do is run CI units 24/7.

Original issue reported on code.google.com by tarmose...@gmail.com on 26 Jun 2012 at 1:46

GoogleCodeExporter commented 9 years ago 
That all is related to FreeHal@HOME Boinc project, not FreeHal itself.

Original comment by tarmose...@gmail.com on 26 Jun 2012 at 1:47

GoogleCodeExporter commented 9 years ago 
It just happened again... settings on the web page just changed from CI to NCI 
and all WU-s started to run at once again. Had to suspend all and them manually 
resume them.

Original comment by tarmose...@gmail.com on 27 Jun 2012 at 10:45

Attachments:

GoogleCodeExporter commented 9 years ago 
And it happened again. After feeder being down for couple of days, it resumed 
working. At the same time Freehal settings just got reverted back to one WU per 
host & NCI. Damn I hate that!
It seems to me, that it happens after every 3-4 days.

It needs fixing ASAP!

Original comment by tarmose...@gmail.com on 4 Jul 2012 at 9:35