Welcome to MilkyWay@home

Machines runs out of units every few minutes

Message boards : Number crunching : Machines runs out of units every few minutes
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Adrian Taylor
Avatar

Send message
Joined: 6 Apr 08
Posts: 13
Credit: 139,088,163
RAC: 0
Message 63002 - Posted: 13 Jan 2015, 10:34:32 UTC

as the title suggests , my only crunching machine will get a few units, crunch them in a minute or so and then sit there for 15 minutes without downloading any more

is there any setting i can add to make it poll for more work more frequently?

when i do a manual "update" then i get another few minutes of work

hope someone can can help :-)

cheers
ID: 63002 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Anthony Ayiomamitis

Send message
Joined: 25 Dec 14
Posts: 8
Credit: 1,007,280
RAC: 0
Message 63009 - Posted: 13 Jan 2015, 18:26:37 UTC

Adrian,

There is a similar thread elsewhere and it seems MW will only allow three tasks per core and irrespective of how slow or fast these will be processed.

See the following: http://milkyway.cs.rpi.edu/milkyway/forum_thread.php?id=3698

Anthony.
ID: 63009 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Adrian Taylor
Avatar

Send message
Joined: 6 Apr 08
Posts: 13
Credit: 139,088,163
RAC: 0
Message 63010 - Posted: 13 Jan 2015, 18:28:22 UTC

yes i understand that limitation but i don't get anything for quite long periods like now for instance :-)

thanks for taking the time to reply
ID: 63010 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Anthony Ayiomamitis

Send message
Joined: 25 Dec 14
Posts: 8
Credit: 1,007,280
RAC: 0
Message 63011 - Posted: 13 Jan 2015, 18:38:11 UTC - in response to Message 63010.  
Last modified: 13 Jan 2015, 18:38:37 UTC

BOINC Manager will always fetch new work after a completed task is uploaded. As a result, you should not have this problem unless no new tasks are available from the server for download.

With my installation, it checks once again a minute later for another fetch in the event the first fetch failed.
ID: 63011 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Chris

Send message
Joined: 13 Apr 12
Posts: 1
Credit: 153,340
RAC: 0
Message 63013 - Posted: 13 Jan 2015, 23:31:33 UTC

Mine asks for new GPU work with every upload, but sometimes the server does not give any. Is the project that starved for work units?

I guess so, the server status shows under 1000 wu available. I'm a bit surprised, thats almost CPDN bare.
ID: 63013 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Adrian Taylor
Avatar

Send message
Joined: 6 Apr 08
Posts: 13
Credit: 139,088,163
RAC: 0
Message 63014 - Posted: 14 Jan 2015, 0:58:18 UTC

I reckon its a combination of a work unit shortage and the very quick turnover in WU...

shame cos i have the fastest rig here and its barely heating the room

;-)
ID: 63014 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Jari Pyyluoma

Send message
Joined: 6 Mar 09
Posts: 4
Credit: 1,027,033,744
RAC: 0
Message 63019 - Posted: 16 Jan 2015, 7:59:24 UTC - in response to Message 63014.  

I see the same problem. It is not a new problem either, it has existed many years. Sometimes it happens because the project releases bad work, that errors out, which makes the server not give out new work. This seems to be the case for me right now. Check your tasks in your account for errors. This project is kind of sloppy, which takes getting used to. Overclocking may also be a culprit regarding errors.
ID: 63019 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Mumak
Avatar

Send message
Joined: 8 Apr 13
Posts: 89
Credit: 517,085,245
RAC: 0
Message 63020 - Posted: 16 Jan 2015, 8:53:59 UTC - in response to Message 63019.  

I'm experiencing the same problem right now. Only getting a few units, sometimes my queues are empty.
ID: 63020 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Herge

Send message
Joined: 11 Oct 09
Posts: 19
Credit: 202,475,569
RAC: 0
Message 63023 - Posted: 16 Jan 2015, 18:24:57 UTC

Not sure if it is the same problem I have. I get about 60 units and when I return them, I don't get any new ones and then I run out of units. So I ended up created a job in Windows Task Scheduler which send an update command every 5 minutes. Since then, the only time I ran out of units was due to the fact that none were available.
ID: 63023 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile mikey
Avatar

Send message
Joined: 8 May 09
Posts: 3315
Credit: 519,950,748
RAC: 21,733
Message 63029 - Posted: 17 Jan 2015, 12:12:41 UTC - in response to Message 63023.  

Not sure if it is the same problem I have. I get about 60 units and when I return them, I don't get any new ones and then I run out of units. So I ended up created a job in Windows Task Scheduler which send an update command every 5 minutes. Since then, the only time I ran out of units was due to the fact that none were available.


I'm not sure you should be saying that out loud...you are essentially tying up the MW Server for YOU and not letting the rest of us in. It's not like each connection doesn't use Server time or the ports we connect on are unlimited, with 21,509 active users you could be the reason I can't get thru sometimes, and THAT'S why the backoff was implemented back in the day. I understand it works for you, but you are sequentially saying 'I am more important that you are and I want MY UNITS and I wan then NOW. That is not what the Project wants though, they want ALL of us to be at least somewhat happy. I'm just saying you want to NOT give away all your secrets in a public forum.
ID: 63029 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
E01902

Send message
Joined: 10 Jan 15
Posts: 5
Credit: 7,320
RAC: 0
Message 63032 - Posted: 17 Jan 2015, 15:21:28 UTC - in response to Message 63029.  

Actually it's important that the project leader(s) hear that people are frustrated and trying to cheat the system. This allows them to manage the cheating. Someones telling them they're calling in extra because xxx.
Now they now why there's an extra load on that server without having to research for hours.
If people can't get enough work they'll move on to other projects. So, either this project has more processing power than it needs, which is a good position to be in, or it has a project problem where it can't turn work around fast enough and the developers need to explain what's going on. Communication from the Project Leader is just as important as developing the code to fix the problem.
Thanks for reminding people of the impact of what they think might be a quick fix, doesn't necessarily help the system if everyone tries the same thing.
ID: 63032 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile mikey
Avatar

Send message
Joined: 8 May 09
Posts: 3315
Credit: 519,950,748
RAC: 21,733
Message 63038 - Posted: 18 Jan 2015, 11:52:06 UTC - in response to Message 63032.  

Actually it's important that the project leader(s) hear that people are frustrated and trying to cheat the system. This allows them to manage the cheating. Someones telling them they're calling in extra because xxx.
Now they now why there's an extra load on that server without having to research for hours.
If people can't get enough work they'll move on to other projects. So, either this project has more processing power than it needs, which is a good position to be in, or it has a project problem where it can't turn work around fast enough and the developers need to explain what's going on. Communication from the Project Leader is just as important as developing the code to fix the problem.
Thanks for reminding people of the impact of what they think might be a quick fix, doesn't necessarily help the system if everyone tries the same thing.


I was careful NOT to say he was 'cheating', he was taking advantage of a Windows feature that let him do things, 'cheating' implies he is doing something illegally, he isn't. What he IS doing is going around the established protocols and potentially causing problems for the rest of us, but this is all due to the Project having problems. If the Project itself could figure out the problems then he wouldn't have been forced to come up with an outside the box way to get units.

ALL I was trying to say was that he might not want to give away all of his secrets as others ARE listening, and they do have the power to affect his 'solution' that NONE of us may like.
ID: 63038 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Mumak
Avatar

Send message
Joined: 8 Apr 13
Posts: 89
Credit: 517,085,245
RAC: 0
Message 63041 - Posted: 18 Jan 2015, 16:30:39 UTC

Hello Admins, anybody here? I'd appreciate a feedback on this issue.
I haven't seen such problems yes, it started a few days ago and persists.
It causes my machines to run dry of GPU units, thus as a side effect the CPU cores allocated to GPUs start CPU tasks, then after a while when a few GPU units are received, they switch back to GPUs and so on. This results in constant suspending/restarting/reshuffling of CPU tasks.
ID: 63041 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : Number crunching : Machines runs out of units every few minutes

©2024 Astroinformatics Group