Welcome to MilkyWay@home

work availability

Message boards : Number crunching : work availability
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · 4 . . . 5 · Next

AuthorMessage
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 6883 - Posted: 27 Nov 2008, 17:07:23 UTC

I think i found out the problem in why the work availability has been so poor. The transitioner was really backed up (probably from running 2 apps instead of 1), so while both assimilators were generating work, it wasn't getting transitioned to a place where you guys could download it fast enough, which only further slowed down the transitioner :P

I took a few steps to speed that process up, so let me know how the work is flowing now.
ID: 6883 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
jedirock
Avatar

Send message
Joined: 8 Nov 08
Posts: 178
Credit: 6,140,854
RAC: 0
Message 6884 - Posted: 27 Nov 2008, 17:22:42 UTC - in response to Message 6883.  

I took a few steps to speed that process up, so let me know how the work is flowing now.

I'm still getting the "No Work Sent" message with a custom application. As I know, my app_info is fine. Does this change affect the new app or just the old one?
ID: 6884 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 6887 - Posted: 27 Nov 2008, 17:46:30 UTC - in response to Message 6884.  
Last modified: 27 Nov 2008, 17:48:13 UTC

I took a few steps to speed that process up, so let me know how the work is flowing now.

I'm still getting the "No Work Sent" message with a custom application. As I know, my app_info is fine. Does this change affect the new app or just the old one?


The assimilator is still generating them... looks like they're being snagged up as fast as they get out. Hopefully once the transitioner catches back up this will go away.
ID: 6887 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Arion
Avatar

Send message
Joined: 10 Aug 08
Posts: 218
Credit: 41,846,854
RAC: 0
Message 6890 - Posted: 27 Nov 2008, 18:14:27 UTC - in response to Message 6887.  

The assimilator is still generating them... looks like they're being snagged up as fast as they get out. Hopefully once the transitioner catches back up this will go away.


Just wanted to say Happy Thanksgiving to you Travis. I can think of a lot of places I would rather be than kicking the server so we can get work. Hope you and yours get to have some quality time this afternoon!


Arion
ID: 6890 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Cori
Avatar

Send message
Joined: 27 Aug 07
Posts: 647
Credit: 27,592,547
RAC: 0
Message 6892 - Posted: 27 Nov 2008, 18:26:20 UTC - in response to Message 6890.  

Just wanted to say Happy Thanksgiving to you Travis. I can think of a lot of places I would rather be than kicking the server so we can get work. Hope you and yours get to have some quality time this afternoon!


Arion

I have to agree... So Happy Thanksgiving to the other side of the pond! ;-)))


Lovely greetings, Cori
ID: 6892 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
John Clark

Send message
Joined: 4 Oct 08
Posts: 1734
Credit: 64,228,409
RAC: 0
Message 6893 - Posted: 27 Nov 2008, 19:01:09 UTC

I would like to echo Arion and Cori's sentiments. Have a good Thanksgiving Day
ID: 6893 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
scruff35

Send message
Joined: 25 Nov 08
Posts: 2
Credit: 21,223
RAC: 0
Message 6907 - Posted: 28 Nov 2008, 1:15:39 UTC

Hi, and Happy Thanksgiving to all my American freinds. Just wondering about wu availablity. I`m still getting "No work sent", even though there are over 5k wu available. Just got one on a forced update, and that`s it. TIA.
ID: 6907 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Alinator

Send message
Joined: 7 Jun 08
Posts: 464
Credit: 56,639,936
RAC: 0
Message 6916 - Posted: 28 Nov 2008, 18:19:47 UTC

One thing to keep in mind about the server status is that it is not updated in real time. Also, even if there appears to be a lot of work available when you look, that can go poof in a hurry! ;-)

Alinator
ID: 6916 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
BarryAZ

Send message
Joined: 1 Sep 08
Posts: 520
Credit: 302,524,931
RAC: 15
Message 6918 - Posted: 28 Nov 2008, 18:49:48 UTC - in response to Message 6916.  

Realizing how difficult it is to keep queues filled, and recognizing that there is something of an artificial set limit on credit per cpu per day, I am a bit confused as to how any given computer can still generate a RAC of in excess of 10000 these days. Heck, there is one dual core AMD 3800 that is showing a current RAC of nearly 25000.

There are also a handful of folks with RAC's of in excess of 100000 and with the current constraints as to hourly credits and available work, I really wonder how that happens.

I mean during the gravy days, I actually did get 100K credits on one day, but these days I have to push things to get 20K. Then again, I don't run single project workstations -- never had.

At one point about a month ago, MilkyWay was generating over 80% of my BOINC RAC, these days it is under 40%. The reduction is due to the change in credit schemes (and hopefully a more rational one will emerge which doesn't punish faster CPU's but which also doesn't provide excess credits to the point of waking up the DA beast), and to the difficulty in keeping work on systems (those less than 2 hour queue limits need to get fixed).


ID: 6918 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Alinator

Send message
Joined: 7 Jun 08
Posts: 464
Credit: 56,639,936
RAC: 0
Message 6919 - Posted: 28 Nov 2008, 19:52:27 UTC

Hmmm...

Agreed, the numbers don't appear to reconcile with what I could see for current performance for the current # 1 host.

I managed to catch some outcome data for it before it got purged, and it was showing about 119 credits being granted for runtimes right around 4000 seconds.

It would seem that to support a RAC of around 34500 you would have to run about 290 tasks per day. At 4 kSecs each, there doesn't seem to be enough seconds in the day to support that, even for a quad!? ;-)

I just took a look at the Computer Summary page for it and it's now showing RAC at over 35000. Of course with insta-purge active, it's pretty hard to get any clear indication of what might be going on from our POV. :-(

I can't even use my own hosts to help me figure this out, since I run them at equal shares with at least 2 other projects and the indications show what I expect to see for RAC given that the basis is high even with the compensation they use here.

Alinator
ID: 6919 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
BarryAZ

Send message
Joined: 1 Sep 08
Posts: 520
Credit: 302,524,931
RAC: 15
Message 6922 - Posted: 28 Nov 2008, 22:53:04 UTC - in response to Message 6919.  

Right, even if you could get all the workunits downloaded and processed (a serious constraint give work unit availability), there are two other constraints in the mix, first, as you noted, how many work units are needed for those 5 digit individual workstation RAC numbers, and second, (according to what Travis posted regarding credits) the project based 'governor' on maximum credits per hour -- such that if you are running a 3G CPU full tilt, you won't get any more credits than if you were running a 1GHz CPU full tilt.

So, there are at least two tactics those ultra high RAC numbers for individual workstations suggest. First, some sort of script which pounds the server to keep the workstation filled with work units (that is something I suspect could be cobbled together without too much effort), and *second*, some capability to *spoof* the server into believing the work is being generated by some 8 or higher core computer so as to not bounce against the 'governor' which Travis has indicated is in place.

Not a big deal to me, I typically exclude outlier RAC numbers from my view of the world. It just something I'm wondering about as I struggle to manually fill queues.


Hmmm...

Agreed, the numbers don't appear to reconcile with what I could see for current performance for the current # 1 host.

I managed to catch some outcome data for it before it got purged, and it was showing about 119 credits being granted for runtimes right around 4000 seconds.

It would seem that to support a RAC of around 34500 you would have to run about 290 tasks per day. At 4 kSecs each, there doesn't seem to be enough seconds in the day to support that, even for a quad!? ;-)

I just took a look at the Computer Summary page for it and it's now showing RAC at over 35000. Of course with insta-purge active, it's pretty hard to get any clear indication of what might be going on from our POV. :-(

I can't even use my own hosts to help me figure this out, since I run them at equal shares with at least 2 other projects and the indications show what I expect to see for RAC given that the basis is high even with the compensation they use here.

Alinator


ID: 6922 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
John Clark

Send message
Joined: 4 Oct 08
Posts: 1734
Credit: 64,228,409
RAC: 0
Message 6926 - Posted: 29 Nov 2008, 9:42:32 UTC
Last modified: 29 Nov 2008, 9:44:58 UTC

Th Server Status page, at 09.32 UTC, says there is 4,089 WUs ready to send, but they are not available to crunchers as new downloads. Yet all the server running status is fine.

The pending credit is as high as I have ever seen it.

Something is preventing this work reaching rigs seeking more work.

I guess one of the server scripts have fallen over and needs a kick?

Am I not the only person with no MW work to crunch? I would guess not!
ID: 6926 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Mr Mystery

Send message
Joined: 21 Nov 08
Posts: 90
Credit: 2,601
RAC: 0
Message 6927 - Posted: 29 Nov 2008, 10:11:19 UTC - in response to Message 6926.  
Last modified: 29 Nov 2008, 10:19:22 UTC

Th Server Status page, at 09.32 UTC, says there is 4,089 WUs ready to send, but they are not available to crunchers as new downloads. Yet all the server running status is fine.

The pending credit is as high as I have ever seen it.

Something is preventing this work reaching rigs seeking more work.

I guess one of the server scripts have fallen over and needs a kick?

Am I not the only person with no MW work to crunch? I would guess not!



Well I'm getting handfuls of stripe nn and test nn work so perhaps your prefs aint correct?
ID: 6927 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
mycal

Send message
Joined: 13 Jan 08
Posts: 19
Credit: 820,482
RAC: 0
Message 6928 - Posted: 29 Nov 2008, 10:18:05 UTC

No problems here both host now on NNT to process present downloads both on 20 WU per CPU limits.

Have not looked at laptop yet.

Activity always run.

Network always available.

Connection time 1.00 days.

Michael
ID: 6928 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Logan
Avatar

Send message
Joined: 15 Aug 08
Posts: 163
Credit: 3,876,869
RAC: 0
Message 6929 - Posted: 29 Nov 2008, 10:26:32 UTC

No problems with downloads here. By the moment...


Best regards.
Logan.

BOINC FAQ Service (Ahora, también disponible en Español/Now available in Spanish)
ID: 6929 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Lord Tedric
Avatar

Send message
Joined: 9 Nov 07
Posts: 151
Credit: 8,391,608
RAC: 0
Message 6930 - Posted: 29 Nov 2008, 10:35:12 UTC

Problems here............
29/11/2008 10:27:26|Milkyway@home|Sending scheduler request: Requested by user. Requesting 488992 seconds of work, reporting 0 completed tasks
29/11/2008 10:27:31|Milkyway@home|Scheduler request completed: got 0 new tasks
29/11/2008 10:27:31|Milkyway@home|Message from server: No work sent
29/11/2008 10:29:41|Milkyway@home|Sending scheduler request: Requested by user. Requesting 489294 seconds of work, reporting 0 completed tasks
29/11/2008 10:29:46|Milkyway@home|Scheduler request completed: got 0 new tasks
29/11/2008 10:29:46|Milkyway@home|Message from server: No work sent
29/11/2008 10:30:42|Milkyway@home|Sending scheduler request: Requested by user. Requesting 489434 seconds of work, reporting 0 completed tasks
29/11/2008 10:30:47|Milkyway@home|Scheduler request completed: got 0 new tasks
29/11/2008 10:30:47|Milkyway@home|Message from server: No work sent
29/11/2008 10:34:07|Milkyway@home|Sending scheduler request: Requested by user. Requesting 489917 seconds of work, reporting 0 completed tasks
29/11/2008 10:34:12|Milkyway@home|Scheduler request completed: got 0 new tasks
29/11/2008 10:34:12|Milkyway@home|Message from server: No work sent

ID: 6930 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile GalaxyIce
Avatar

Send message
Joined: 6 Apr 08
Posts: 2018
Credit: 100,142,856
RAC: 0
Message 6931 - Posted: 29 Nov 2008, 10:39:48 UTC


I have more than enough to crunch on, I can't crumble ;)

And a RAC I never would have expected to get. Crikey!


ID: 6931 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
WimTea

Send message
Joined: 16 Nov 07
Posts: 23
Credit: 4,774,710
RAC: 0
Message 6932 - Posted: 29 Nov 2008, 10:51:46 UTC

Couple of days now my (few) comps get only a handful every now and then.
Are there more available with the test app? If so I'm going to switch over to that one.
ID: 6932 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Conan
Avatar

Send message
Joined: 2 Jan 08
Posts: 122
Credit: 69,479,962
RAC: 1,426
Message 6933 - Posted: 29 Nov 2008, 10:55:43 UTC
Last modified: 29 Nov 2008, 10:56:20 UTC

Well with over 7,500 available work units and I am getting the message that 'No work sent', even though I have just 3 work units on one 4 core machine and 5 work units on another 4 core machine (the one with 5 also got the message that 'Not requesting new work', which I was requesting and still want).

Perhaps they are all 'test' work units?, which I am not processing at the moment.
ID: 6933 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile caspr
Avatar

Send message
Joined: 22 Mar 08
Posts: 90
Credit: 501,728
RAC: 0
Message 6934 - Posted: 29 Nov 2008, 11:21:04 UTC - in response to Message 6933.  

Well with over 7,500 available work units and I am getting the message that 'No work sent', even though I have just 3 work units on one 4 core machine and 5 work units on another 4 core machine (the one with 5 also got the message that 'Not requesting new work', which I was requesting and still want).

Perhaps they are all 'test' work units?, which I am not processing at the moment.



Over 9k ready to send now but the transitioner backlog is at 3hrs and pendings are going through the roof!
A clear conscience is usually the sign of a bad memory



ID: 6934 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
1 · 2 · 3 · 4 . . . 5 · Next

Message boards : Number crunching : work availability

©2024 Astroinformatics Group