One-time big cache wanted, but refused?
19 Oct 2006 20:46:14 UTC
Topic 191960
(moderation:
Hi,
As I'm going on a holiday, I wanted to fill the cache of a comp with a week of Einsteins (it's not networked the coming week). The comp is attached to some other projects, but they're all on 'no new work' as their deadlines are a week maximum.
I got the cache filled with 5 days of work (using BoincStudio), but after a request for another 3 days, Einstein tells me 'No More':
19/10/2006 22:33|Einstein@Home|Requesting 172800 seconds of new work
19/10/2006 22:33|Einstein@Home|Scheduler request succeeded
19/10/2006 22:33|Einstein@Home|Message from server: No work sent
19/10/2006 22:33|Einstein@Home|Message from server: (won't finish in time) Computer on 91.5% of time, BOINC on 98.8% of that, this project gets 5.9% of that
19/10/2006 22:33|Einstein@Home|No work from project
Anyone got an idea how to bypass this? Apart from detaching from projects or changing resourceshares, that is... Increasing the number of CPUs reported didn't help...
TIA,
WimTea
Make tea, not coffee !
One-time big cache wanted, but refused?
)
Hmmm, we're a little thin on info here. I'm assuming this is the Pentium M?
The reason for the refusal right now is pretty simple. Based on what the EAH servers know about this host up to now, you appear to be carrying ~21 days worth of work, and we don't have a clue how much work is left onboard for the other projects.
In any event, until all the other project work is done I don't think there is a way to force more work for EAH right now without increasing the resource share temporarily to do it.
Alinator
You can suspend the other
)
You can suspend the other projects BOINC will think Einstein is the only project on that host and should get more work. This is NOT a recommended practice.
BOINC WIKI
BOINCing since 2002/12/8
Thanks for the replies, Keck
)
Thanks for the replies, Keck and Alinator.
It was indeed the Pentium M that could use the extra work. At the end it had nothing to crunch for 1.5 days :-( .
I realize now I've been a bit daft in just NNW'ing all other projects, it would never done the trick. Instead I should have been suspending them, so BOINC would indeed have thought Einstein to be the only 'active' project. Well, it was a lesson learned.
Wim.
[edit] I agree it is not a reccommended practise, but it is a nice workaround when you mostly have projects with deadlines of 1 week (and are gone over 1 week), which Einstein fortunately has not.
Make tea, not coffee !
No problemo! :-) I took a
)
No problemo! :-)
I took a look at your overall at BoincStats, and it doesn't look like you took too big a hit, so I guess it's a case of "No harm, no foul". ;-)
Also, as you said it would most likely worked out better to just suspend the others, increase your CI, update EAH, and then reset everything back to it's normal setting right before you left. Although, you still most likely would have had to NNW the short deadline projects early to account for where the deadlines for onboard work were in the vacation time interval so you could let them run out and report before you left.
Alinator