Message boards :
Number crunching :
Not getting much work
Message board moderation
Author | Message |
---|---|
Send message Joined: 7 Apr 15 Posts: 13 Credit: 136,390,419 RAC: 0 |
For much of the last two days, my Linux hosts have not been receiving much work. Even though the project indicates it has work, the common response to work requests is to simply receive 0 tasks. Is there a problem with availability of work for Linux? Or even in general? I don't have many Windows hosts, but they are also not getting much work when requested. Also, in reviewing my tasks that are "Pending" status, many of them report the matching WU is "Unsent" so there really should be work, but it seems it's not going out. |
Send message Joined: 6 Mar 15 Posts: 28 Credit: 16,721,329 RAC: 0 |
...Even though the project indicates it has work, the common response to work requests is to simply receive 0 tasks... You are not the only one. I've been seeing the same thing on all my Windows hosts... :-( |
Send message Joined: 7 Apr 15 Posts: 13 Credit: 136,390,419 RAC: 0 |
Now, as of about 30 minutes ago, I've started receiving messages after the work request "no applications" so hopefully that means we have something being worked on server side. :) |
Send message Joined: 6 Mar 15 Posts: 28 Credit: 16,721,329 RAC: 0 |
My latest attempts to get work still get this: 1572 Universe@Home 8/22/2015 12:18:38 PM Sending scheduler request: To fetch work. 1573 Universe@Home 8/22/2015 12:18:38 PM Requesting new tasks for CPU 1574 Universe@Home 8/22/2015 12:18:39 PM Scheduler request completed: got 0 new tasks 1575 Universe@Home 8/22/2015 12:18:39 PM No tasks sent I've turned on the work_fetch_debug logs, but I don't think that is telling me anything about the server side. |
Send message Joined: 6 Mar 15 Posts: 28 Credit: 16,721,329 RAC: 0 |
I've been getting work off and on since about 1900 UTC. |
Send message Joined: 7 Apr 15 Posts: 13 Credit: 136,390,419 RAC: 0 |
Well, over the last 8 hours, things have slowly improved. Still have a slow trickle coming in, but also still not at a complete cache, even considering the current limit of 8 WU's per core. At least some of my hosts finally have more work on hand than cores. :) |
Send message Joined: 6 Mar 15 Posts: 28 Credit: 16,721,329 RAC: 0 |
...considering the current limit of 8 WU's per core... It seems to me that's either not being enforced or isn't being implemented properly. In addition to other hosts of mine exceeding that value, my 4-thread i5 has 57 tasks in progress right now. |
Send message Joined: 4 Feb 15 Posts: 846 Credit: 144,180,465 RAC: 0 |
I'm working on WU's sending process, please be patient, things going better ;) Krzysztof 'krzyszp' Piszczek Member of Radioactive@Home team My Patreon profile Universe@Home on YT |
Send message Joined: 7 Apr 15 Posts: 13 Credit: 136,390,419 RAC: 0 |
Correct. Overnight some of my team mates saw the message that they'd hit the limit of tasks in progress of 8 per core. They also have indicated that limit seems to have since been lifted. :) |
Send message Joined: 7 Apr 15 Posts: 13 Credit: 136,390,419 RAC: 0 |
Thanks for the Post Krzysztof. It always helps when the management lets the volunteers know things are going on behind the curtain. :) |
Send message Joined: 7 Apr 15 Posts: 13 Credit: 136,390,419 RAC: 0 |
Well, I'm still waiting for things to get better. Way to many cores that aren't getting work here. Looking for ideas. Linux (Ubuntu 14.04) Boinc (7.4.25) Min cache (I've tried .5, 1, 2 and 5) Additional Days (I've tried 1.25, 2, 3 and 10) I've removed most other projects and have taken the 2 left and set them to NNT and even Suspended them. Still not getting work. Yet, when I enable work fetch on one of those projects (SRBase), I get tons. I've tried adjusting the project specific dcf from the current calculated value of 1.00000, to 0.10000 with no improvement either. One thing that may be unique to my world is that almost all of my hosts are hitting the project using a single firewall source address. I'm sure I've got more coming at you than your average User. :) Just some sample work requests from a single representative host: Tue 25 Aug 2015 08:02:32 AM CDT | Universe@Home | [work_fetch] set_request() for CPU: ninst 12 nused_total 11.00 nidle_now 0.00 fetch share 1.00 req_inst 1.00 req_secs 10104408.83 |
Send message Joined: 4 Feb 15 Posts: 846 Credit: 144,180,465 RAC: 0 |
Mumps, currently on server works work generator and it generate more than 1kk WU. This make server slow and daemons delayed... Things should fix itself in about 50 hours (I know, it's a lot)... Krzysztof 'krzyszp' Piszczek Member of Radioactive@Home team My Patreon profile Universe@Home on YT |
Send message Joined: 13 May 15 Posts: 87 Credit: 4,320,738 RAC: 0 |
Mumps, currently on server works work generator and it generate more than 1kk WU. This make server slow and daemons delayed... Things should fix itself in about 50 hours (I know, it's a lot)... I'm curious. This 'Got 0 New Tasks'... is BOINC suppose to send why there is no work or is this something that the project has to configure so BOINC can tell the user? I wonder because Universe is not the only project that does that... |
Send message Joined: 21 Feb 15 Posts: 52 Credit: 318,272 RAC: 0 |
I'm curious. This 'Got 0 New Tasks'... is BOINC suppose to send why there is no work or is this something that the project has to configure so BOINC can tell the user? I wonder because Universe is not the only project that does that... Same here, "got 0 new task". |
Send message Joined: 7 Apr 15 Posts: 13 Credit: 136,390,419 RAC: 0 |
Agreed. Many other projects do actually indicate when they have no work available. As an example, here's a work request from Sztaki, a project that commonly doesn't have work available at all times of the day. Thu 27 Aug 2015 08:10:48 AM CDT | SZTAKI Desktop Grid | update requested by user Thu 27 Aug 2015 08:10:50 AM CDT | SZTAKI Desktop Grid | Sending scheduler request: Requested by user. Thu 27 Aug 2015 08:10:50 AM CDT | SZTAKI Desktop Grid | Requesting new tasks for CPU Thu 27 Aug 2015 08:10:52 AM CDT | SZTAKI Desktop Grid | Scheduler request completed: got 0 new tasks Thu 27 Aug 2015 08:10:52 AM CDT | SZTAKI Desktop Grid | No work sent Thu 27 Aug 2015 08:10:52 AM CDT | SZTAKI Desktop Grid | (Project has no jobs available) But I don't know if they need to set something up in any of the server side configuration files to do so. |
Send message Joined: 6 Mar 15 Posts: 28 Credit: 16,721,329 RAC: 0 |
Krzysztof Piszczek wrote: ... Things should fix itself in about 50 hours... Things still seem a bit gummed up almost 96 hours later. Oh, well... Plenty of other projects to crunch while patiently waiting. :-) |
Send message Joined: 7 Apr 15 Posts: 13 Credit: 136,390,419 RAC: 0 |
Yup. Really gummed up if the steadily decreasing number of WU's in progress is any indicator. I personally don't have anything left. I'll hang around for when the log-jam breaks free, but I'm forced to work towards goals elsewhere in the interim. :) |
Send message Joined: 13 May 15 Posts: 87 Credit: 4,320,738 RAC: 0 |
Yup. Really gummed up if the steadily decreasing number of WU's in progress is any indicator. I personally don't have anything left. I'll hang around for when the log-jam breaks free, but I'm forced to work towards goals elsewhere in the interim. :) I can't be 100% sure, but from last night to this morning, the available WUs hasn't budged. I was having problems getting WUs at least yesterday. I did a reset then and got a WU and then nothing more even when I was almost out of WUs. Tried reset again, nada. |
Send message Joined: 13 May 15 Posts: 87 Credit: 4,320,738 RAC: 0 |
Just checked BOINC, got 3 new Universe WUs sitting there waiting to crunch. |
Send message Joined: 21 Feb 15 Posts: 8 Credit: 2,270,934 RAC: 0 |
I just had to abort some because there was no way I could get them crunched in time. |