Advanced search

Message boards : News : "No new work/too much work" problems

Author Message
ignasi
Send message
Joined: 10 Apr 08
Posts: 254
Credit: 16,836,000
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwat
Message 20920 - Posted: 13 Apr 2011 | 11:09:12 UTC

We have found what prevented most of users to receive work despite having WUs in queue. Everybody should be receiving work now. There's plenty of WUS to crunch!

ftpd
Send message
Joined: 6 Jun 08
Posts: 152
Credit: 328,250,382
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 20924 - Posted: 13 Apr 2011 | 11:39:14 UTC - in response to Message 20920.

Ignasi,

I just received 8 (eight) wu's for 1 (one) machine with gtx480-card.

Strange????
____________
Ton (ftpd) Netherlands

Toni
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 9 Dec 08
Posts: 1006
Credit: 5,068,599
RAC: 0
Level
Ser
Scientific publications
watwatwatwat
Message 20925 - Posted: 13 Apr 2011 | 11:55:49 UTC - in response to Message 20924.

What's your "additional work buffer"? We had to temporarily change the scheduler algorithm.

ftpd
Send message
Joined: 6 Jun 08
Posts: 152
Credit: 328,250,382
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 20926 - Posted: 13 Apr 2011 | 13:34:07 UTC - in response to Message 20925.

Toni,

Normally i receive max. 2 wu's for this machine with gtx480. ID 35174

Enough info?
____________
Ton (ftpd) Netherlands

Toni
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 9 Dec 08
Posts: 1006
Credit: 5,068,599
RAC: 0
Level
Ser
Scientific publications
watwatwatwat
Message 20927 - Posted: 13 Apr 2011 | 14:56:04 UTC - in response to Message 20926.

Can you please cancel some of the non-running ones and see if you get one more?

ftpd
Send message
Joined: 6 Jun 08
Posts: 152
Credit: 328,250,382
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 20929 - Posted: 13 Apr 2011 | 16:17:06 UTC - in response to Message 20927.
Last modified: 13 Apr 2011 | 16:21:07 UTC

Toni,

I cancelled all 8 wu's.

Downloading 6 new ones including 1 acemd2 (small wu) which is not in prerefence.

After 30 seconds again 3 new ones including 1 acemd2.

So, something is very wrong at the moment for this machine (gtx480).

I also downloaded 3 new wu's acemd2 for my gtx295 machine, which is OK. (max=4).

After some time (10 minutes) download for this machine extra 2 wu's, which is not OK now!

Enough info?

Good luck!
____________
Ton (ftpd) Netherlands

Toni
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 9 Dec 08
Posts: 1006
Credit: 5,068,599
RAC: 0
Level
Ser
Scientific publications
watwatwatwat
Message 20935 - Posted: 13 Apr 2011 | 16:50:52 UTC - in response to Message 20929.
Last modified: 13 Apr 2011 | 16:51:09 UTC

For now we need everybody to get some WUs. We'll keep debugging this issue. Thanks for reporting. Please let us know if by any chance the situation fixes by itself.

ftpd
Send message
Joined: 6 Jun 08
Posts: 152
Credit: 328,250,382
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 20936 - Posted: 13 Apr 2011 | 17:46:04 UTC - in response to Message 20935.

Toni,

Of course i will do that!

Good luck!
____________
Ton (ftpd) Netherlands

ftpd
Send message
Joined: 6 Jun 08
Posts: 152
Credit: 328,250,382
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 20940 - Posted: 13 Apr 2011 | 19:12:27 UTC - in response to Message 20936.
Last modified: 13 Apr 2011 | 19:13:10 UTC

Toni,

I have one computer xp-pro 64bits with gtx260 card.
Just downloaded 6 wu's (long).

Prerefence is long wu 's and 10 days buffer (for ALL my machines).
Normally download was 2 wu's max.

Good luck!
____________
Ton (ftpd) Netherlands

Otis11
Send message
Joined: 2 Aug 09
Posts: 21
Credit: 197,088,189
RAC: 0
Level
Ile
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 20946 - Posted: 14 Apr 2011 | 2:45:50 UTC

I'm running 2 x 260s with the preference for only long WUs but now have 12 WUs (10 long 2 ACEMD2)

Cache set for 7 days but I don't need that cache for GPUGrid tasks... is this getting set back to max 4 or do I need to tweak my end?

Anoobis
Send message
Joined: 9 Dec 10
Posts: 2
Credit: 1,557,220
RAC: 0
Level
Ala
Scientific publications
watwatwatwatwatwatwat
Message 20950 - Posted: 14 Apr 2011 | 7:59:49 UTC

I have begun receiving WUs, but they are Long-Runs, and I have those disabled. I CAN NOT stand the time limits for long runs! I end up getting 14 hours into 16 and the limit is over, wasting my time and money. I don't run this 24/7 or even every day.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 20951 - Posted: 14 Apr 2011 | 9:04:22 UTC - in response to Message 20950.

The server seems to ignore it for some reason.
We are trying to fix it.

gdf

Toni
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 9 Dec 08
Posts: 1006
Credit: 5,068,599
RAC: 0
Level
Ser
Scientific publications
watwatwatwat
Message 20954 - Posted: 14 Apr 2011 | 9:54:03 UTC - in response to Message 20950.
Last modified: 14 Apr 2011 | 9:55:57 UTC

Anoobis: probably you have the "allow non-preferred apps" on. Try disabling it.

Toni
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 9 Dec 08
Posts: 1006
Credit: 5,068,599
RAC: 0
Level
Ser
Scientific publications
watwatwatwat
Message 20956 - Posted: 14 Apr 2011 | 10:03:38 UTC - in response to Message 20954.

Ton: does decreasing the work buffer help? The new scheduler setting may be pushing stuff in the work buffer more than the old one.

Richard Haselgrove
Send message
Joined: 11 Jul 09
Posts: 1576
Credit: 5,607,036,851
RAC: 8,626,830
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 20958 - Posted: 14 Apr 2011 | 10:40:20 UTC

What brought this problem on in the first place? Have you been updating the BOINC server code? If so, http://boinc.berkeley.edu/trac/changeset/23360/ may be relevant.

Profile ~killer~
Avatar
Send message
Joined: 27 Jan 11
Posts: 5
Credit: 44,264,057
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 20959 - Posted: 14 Apr 2011 | 11:56:50 UTC
Last modified: 14 Apr 2011 | 12:00:43 UTC

Finally it happened!
Finally, you can download the job a few days ahead, in case the server does not allow new tasks.
Up to this point has always loaded a maximum of 2 assignments.
Please do not remove this possibility.
At the moment, I managed to get 10 jobs, which will run about 3 days.

Kirby54925
Send message
Joined: 21 Jan 11
Posts: 31
Credit: 70,061,988
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 20965 - Posted: 15 Apr 2011 | 3:54:45 UTC

Well, the situation seems to have swung the other way: there is no work at all for both the acemdlong and the acemd2 apps.

Otis11
Send message
Joined: 2 Aug 09
Posts: 21
Credit: 197,088,189
RAC: 0
Level
Ile
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 20967 - Posted: 15 Apr 2011 | 4:52:12 UTC - in response to Message 20965.

Well, the situation seems to have swung the other way: there is no work at all for both the acemdlong and the acemd2 apps.


This is exactly why they had the low caches and short turnarounds... they have a limited number of WUs and need the current ones to make the next batch. Because it's all sitting in people queues, there is less work for us to do.

Just let it ping the server for a few minutes and as soon as someone turns something in you should be able to grab one.

Hope they fix this soon. In the mean time, lower your cache to .75 days to keep it about where it was before. With that you should have plenty of time to get the next WU before you complete the current one yet not stock up and make other people idle. If enough people do this we'll get back to normal operation until they fix this on the server side.

Kirby54925
Send message
Joined: 21 Jan 11
Posts: 31
Credit: 70,061,988
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 20968 - Posted: 15 Apr 2011 | 5:37:43 UTC - in response to Message 20967.

I've always set it to 0.02 days. That way, it only gets new WUs when the current one is about half an hour away from finishing.

Toni
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 9 Dec 08
Posts: 1006
Credit: 5,068,599
RAC: 0
Level
Ser
Scientific publications
watwatwatwat
Message 20969 - Posted: 15 Apr 2011 | 7:23:33 UTC

The scheduler change we attempted sent out all the WUs. We'll try to fix today.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 20971 - Posted: 15 Apr 2011 | 8:06:13 UTC - in response to Message 20969.

it is reverted to the old policy for now.

gdf

Anoobis
Send message
Joined: 9 Dec 10
Posts: 2
Credit: 1,557,220
RAC: 0
Level
Ala
Scientific publications
watwatwatwatwatwatwat
Message 20979 - Posted: 15 Apr 2011 | 15:51:59 UTC
Last modified: 15 Apr 2011 | 15:52:34 UTC

I have the "Accecpt 3rd party Apps" thing set, so that miiiight be the problem. I also have my work load set at 0.4. I still have what I believe to be the latest BOINC, unless a new one came out in the last couple months. I am getting no work at all except 3rd party apps, and even those are rare. I have been putting my GPU to work for Einstein@Home 100% for the time being, but it would be nice to have more (even 3rd party non long-run) from this project.

This GTX470 SOC is stellar, but I wish I waited 3 months for the GTX 560... The 470 frequently reaches sterilization temperature :D

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 20988 - Posted: 17 Apr 2011 | 10:54:39 UTC - in response to Message 20979.

Hi Anoobis ... please open a post over in the "Number Crunching" forum so we can help you get your 470 producing the best it can. You could start by setting Einstein to NNT. GPUGrid should get a WUs before Einstein completely dries up. If not press the update button and keep trying as sometimes the scheduler has problems determining if you have a reliable PC or not.
____________
Thanks - Steve

Ungelovende
Send message
Joined: 13 Jan 08
Posts: 1
Credit: 107,960,836
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 20989 - Posted: 17 Apr 2011 | 17:54:51 UTC - in response to Message 20988.
Last modified: 17 Apr 2011 | 18:36:36 UTC

I'm trying to get work for this computer:
http://www.gpugrid.net/show_host_detail.php?hostid=90459

My gpugridconfig is set up to accept any WU.

getting this:
18.04.2011 02:44:59 | GPUGRID | update requested by user
18.04.2011 02:45:03 | GPUGRID | Sending scheduler request: Requested by user.
18.04.2011 02:45:03 | GPUGRID | Requesting new tasks for NVIDIA GPU
18.04.2011 02:45:05 | GPUGRID | Scheduler request completed: got 0 new tasks
18.04.2011 02:45:05 | GPUGRID | No work sent
18.04.2011 02:45:05 | GPUGRID | ACEMD beta version is not available for your type of computer.

EDit: did get some WU's after several manual updates :-)

Post to thread

Message boards : News : "No new work/too much work" problems

//