| Info | Message |
|---|---|
| 41) Message boards : Questions and problems : BOINC caused small catastrophe
Message 71877 Posted 25 Aug 2016 by Paul Schauble |
I just had BOINC case a small catastrophe. I'll describe what happened and hope that the behavior can be changed. The main machine on which I run BOINC has two problems that have required workarounds. The machine has an AMD Phenom II CPU and runs Windows 10 64 bit. 1. The machine is the main machine for backups. And, due to some hardware glitch, it is not possible to run GPU computation while doing a backup. If yo do, the SCSI controller will crash, requiring a power off reboot and destroying the data on whatever tape is in use. The workaround, of course, is to set BOINC to suspend GPU use when the backup program is running. Problem solved. 2. I'm in Phoenix, where summer temperatures of 115F or so are not uncommon. The air conditioning cannot keep up. So in the summer I want to limit hours that GPU computations run to exclude the hottest part of the day. The workaround here is to suspend GPU in BOINC and activate a Windows scheduled task that at the right time of day runs the command "C:\Program Files\BOINC\boinccmd.exe --set_gpu_mode always 43200" to restart GPU computation for a time. This has also been working fine. This evening, my backup hung and as I discovered, ruined the tape set being updated. A quick look at BOINC showed that it was running GPU tasks, despite the running backup program being set as a GPU exclusive program. I can only speculate what happened, but here goes. Clearly the set_gpu_mode command overrides the menu setting to suspend GPU computations exactly as documented. It appears that his command ALSO overrides the specification of a gpu exclusive application. Otherwise, why was BOINC running GPU tasks while the backup program was running? If I'm right, I think this should be fixed. If nothing else, the set_gpu_mode is not documented to override "suspend GPU when.." settings and I think it should not. Thanks for listening, ++PLS |
| 42) Message boards : Questions and problems : Resource share apparently does nothing at all
Message 67380 Posted 30 Jan 2016 by Paul Schauble |
I'm currently running BOINC 7.6.22(x64) but this problem goes back to the scheduler rewrite a while back. I'm running Seti (resource share 250), Einstein (RS 2), and 3 other projects with resource share 50. I have not run Einstein for a very long time but it remains in my project list. Since Seti does not currently have a GPU version on 1/16 I decided to enable Einstein. Since 1/19, my computer has not run a single Seti task. I've seen this before. Whenever I enable Einstein, it completely takes over my machine and almost nothing else runs. I've tried medium resource shares for Einstein, I've tried a share of 1 and 2, and everything give this same result: almost nothing else runs. This is why I gave up running Einstein. I think there is a scheduler problem here that should be fixed. |
| 43) Message boards : Questions and problems : BOINC running way too many processes
Message 64228 Posted 14 Sep 2015 by Paul Schauble |
That may be it. Recalling that when stepping back to 7.4.42, I still had the same problem. I uninstalled 7.4.42 and went through the registry and the file system and deleted any references to BOINC or any of its components that I found. I reinstalled 7.4.42 and it is now working correctly. It may be noteworthy that I do not presently have a cc_config.xml file at all. I find "allow_multiple_clients" a very confusing name. Does this mean it allows multiple boinc.exe processes or does it allow multiple boincmgr.exe clients? It might be nice to rename this to make it clear. Also, where do I get the skeleton cc_config.xml file? I would have though that reinstalling 7.4.42 would have installed one in the data directory. It did not. Thanks |
| 44) Message boards : Questions and problems : BOINC running way too many processes
Message 64142 Posted 9 Sep 2015 by Paul Schauble |
Well, this is just disgusting. I went back to 7.4.42 and have exactly the same problem. - Starting boing manager starts two boinc processes, both children of boincmgr; - and each boinc process has 4 seti processes as children. - exiting boing manager with the stop processing box checked does not stop ANY boinc.exe processes, so two are left running, each with 4 seti child processes. - restarting boinc manager starts 2 more boinc child processes, each with 4 stei processes. This isn't useful to run. What can I do now? |
| 45) Message boards : Questions and problems : BOINC running way too many processes
Message 64140 Posted 9 Sep 2015 by Paul Schauble |
BTW, killing processes is pretty iffy. Killing a seti process flatly does not work. BOINC just fails the task and starts a new one. Killing a boinc.exe will kill the seti processes it started, but if I kill the wrong one the manger will just restart it. Of the two boinc.exe processes, one is controlled by the boinc manager and the other is just sitting there running. ++PLS |
| 46) Message boards : Questions and problems : BOINC running way too many processes
Message 64139 Posted 9 Sep 2015 by Paul Schauble |
I have now uninstalled and reinstalled BOINC 7.6.9 and rebooted numerous times. Each time I start BOINC I get two BOINC.exe processes and 8 SETI processes. It is completely consistent, it happens every time. Is there a place where I can download an older version that might not have this bug? Thanks, ++PLS |
| 47) Message boards : Questions and problems : BOINC running way too many processes
Message 64034 Posted 2 Sep 2015 by Paul Schauble |
I am running SETI@Home CPU only. Full configuration is at the end of this message. I updated to BOINC 7.6.6 Monday and 7.6.9 today. Considering only the 7.6.9 version, I have a 4 core CPU. Starting the BOINC manager starts 8 SETI processes and 2 boinc.exe processes. The SETI processes each get about 10% of the available CPU. BOINC manager shows 4 running tasks. Exiting the BOINC manager with the "Stop running" option set leaves all of this running. Restarting BOINC manager now has 4 boinc.exe processes and 16 SETI processes running. I will revert to 7.4.42 for this time being. Configuration: BOINC version 7.6.9 Windows 7, 64 bit CPU AMD Phenom 4 core. CPU Nvidia GTS 450 |
| 48) Message boards : Questions and problems : What is the currently recommended version of VirtualBox?
Message 62463 Posted 7 Jun 2015 by Paul Schauble |
It would be very nice if the BOINC home page showed the current BOINC version and the recommended VirtualBox version. |
| 49) Message boards : Questions and problems : Puzzling allocaton of work
Message 61917 Posted 26 Apr 2015 by Paul Schauble |
I know about the difference in credit paid. But right now I'm running almost all Einstein tasks with only a very occasional SETI task. I want it the other way around. I'll try setting the resource share to 0 and see what happens. How does the resource share business work when there are multiple resources (CPU, GPU) used differently by various tasks? |
| 50) Message boards : Questions and problems : Puzzling allocaton of work
Message 61916 Posted 26 Apr 2015 by Paul Schauble |
Cache is 2 days + 2 days. I wish it were possible to set them by project as this is too short for SETI's long outages. |
| 51) Message boards : Questions and problems : Puzzling allocaton of work
Message 61864 Posted 25 Apr 2015 by Paul Schauble |
The more I watch Boinc with the new scheduler, the less I understand what it's doing. I run Einstein and Seta as GPU tasks and run a couple of other projects that provide strictly CPU tasks. With the resource allocation I had been using, Einstein was running an average credit around 12,000 and SETI an average credit around 4,000. This wasn't what I wanted, I wanted SETI to be the primary task and basically only wanted Einstein to run when SETI was out of tasks. So I started dropping the resource share for Einstein. At the start of April, I set it to 1 (0.28%). SETI is set to 100 (28.5%). So I am expecting that 100 times more GPU will be used on SETI tasks than on Einstein tasks. Now, almost a month later, Einstein is running a average credit of 17,000, SETI is running at 1340, and I have will over 100 hours of Einstein tasks due by 5/1. Let me get this straight. Is reducing the "Resource Share" supposed to cause a project to run more tasks? What do I need to do to reduce the number of Einstein tasks run? Shut the project off entirely? What is the BOINC scheduler actually doing here? ++PLS |
| 52) Message boards : Questions and problems : Want to contribute to source code
Message 60662 Posted 4 Mar 2015 by Paul Schauble |
Where does one find the source code? ++PLS |
| 53) Message boards : Questions and problems : BOINC not downloading tasks as directed
Message 60661 Posted 3 Mar 2015 by Paul Schauble |
I now have my cache set to 2 days plus 1 day and the cached workload is pretty much down to that. I'm now in a situation where my Seti, my highest priority project has no tasks; the Seti server has tasks. Einstein, my lowest priority project, has a large number of tasks cached and continues to download more. When communicating with Seti, BOINC is saying SETI@home | Not requesting tasks: don't need (CPU: job cache full; NVIDIA GPU: job cache full) I've wondered whether the cache settings were per project or for all projects in total. It seems to be for all project together. I guess the people who designed the scheduling algorithm didn't know about priority inversion. The scheduler has filled the cache completely with low priority tasks so that there is no room to download high priority ones. This is really ugly. |
| 54) Message boards : Questions and problems : BOINC not downloading tasks as directed
Message 60536 Posted 25 Feb 2015 by Paul Schauble |
Would a large cache account for such a low total amount of work? The larger the cache you ask for, the less work is actually cached? Actually, the problem was different. I has set all projects to no new work a while back so I could shut down the machine while away. When I set the projects back to accept new work, only a small amount was downloaded. Thus this thread. I just discovered that, somehow, the local settings for work cached both had been set to .1 days. No idea how that happened. I know I didn't do it. When I hit the clear button in Computing Preferences, the settings became 5 and 5 as set in the web interface. I also discovered that using the web interface to set the cache to 3 and 2 will not be picked up by BOINC until I hit the clear button again. Apparently when the 5,5 setting was picked up it became another local override. The setting to 3,2 was never pick up even after the project it was set on did an update AND after I forced an update several times. Is this the intended behavior? It seems wrong. I'll run for a while with 3,2 and see what that looks like. I can already see that the task downloaded when the settings was 5,5 are enabling a BOINC behavior that a really strongly dislike. The previous schedule had a strong tendancy with my project list to run one task from each project. The new schedule has a very strong tendancy to run 4 (on a 4 core system) tasks from the same project, thens shift to 4 tasks from a different project. The problem is that Cosmology tasks are very large and my machine can't run 4 at once and still be usable. I much prefer the old behavior. |
| 55) Message boards : Questions and problems : BOINC not downloading tasks as directed
Message 60496 Posted 23 Feb 2015 by Paul Schauble |
Configuration: BOINC 7.4.36, Windows XP, AMD 4 core CPU, nvidia GTS 450S I have the computer preferences set to store 5 days work and store up to an additional 5 days work. This is to cover the long outages that Seti@home has. Right not, I have a total of 1.5 days work queued. Every project I see in the event log sending update request is saying SETI@home | Not requesting tasks: don't need (CPU: not highest priority project; NVIDIA GPU: not highest priority project) Why aren't more work units being downloaded? My project list is Cosmology@Home, Einstein@Home, Enigma@Home, LHC@Home, SETI@Home, VirtualLHC@Home. A couple of these have very large numbers of work units available, but BOINC isn't trying to get anything. Why? Thanks, ++PLS |
| 56) Message boards : Questions and problems : "--fetch_minimal_work" really fetches 1 work unit per project .. expected?
Message 49623 Posted 15 Jun 2013 by Paul Schauble |
That's not a very useful definition of "device" with respect to CPU cores. Everything else in BOINC schedules by core, not by CPU. ++PLS |
| 57) Message boards : Questions and problems : Comment on BOINC 7
Message 46459 Posted 24 Nov 2012 by Paul Schauble |
After running BOINC 7 for a couple of months, I have to make one comment on the new design. I don't like it. I am subscribed to 7 projects, including Cosmology@Home. My machine has 4 CPU cores and runs 32 bit Windows. The previous version of BOINC would keep a workunit inventory for most of the projects. I would usually run one Cosmology and 3 workunits from other projects. This worked fine. The new BOINC tends to download a large batch of workunits for one project and run nothing only that one project until the batch is done. Cosmology work units are huge and one cannot reasonably run 4 Cosmology work units at one time on a 32 bit system. They take too much memory. So everything works fine, except when the scheduler gets its large batch of Cosmology. For the next few days my machine is very slow doing anything else. I liked the old way better. |
| 58) Message boards : BOINC client : Work fetch policy
Message 43750 Posted 24 Apr 2012 by Paul Schauble |
Shouldn't the work fetch policy guarantee that each active project always has at least one work unit? I notice that BOINC 7 tends to fetch a large bunch of work units for Cosmology, which has long running work units. This allows other CPU projects to finish all their WUs, then sit idle for several days before the number of Cosmology WUs declines to the point where anything fetches will happen on the other projects. I don't think this is right. ++PLS |
| 59) Message boards : Questions and problems : Can't run BOINC from non-admin user
Message 42830 Posted 2 Mar 2012 by Paul Schauble |
Boinc 6.12.34 Windows XP SP3 current in updates After running Boinc with protected execution and several projects without problem for a year, in order to run the new LHC Test4Theory. I uninstalled Boinc, then reinstalled it from and admin user and had protected execution unchecked and let any use manage checked. Now, Boinc seems to run fine from the admin account. But from any other user, starting the Boinc Manager give me a message that the user is not authorized and I need to get an admin to add the user to the boinc_users group. Well, this user is already a member of that group. I have tried deleting and reinstalling, and then repairing without success. How do I fix this to allow any user to run the Boinc manager? ++PLS |
| 60) Message boards : BOINC client : .Net wrapper??
Message 40896 Posted 30 Oct 2011 by Paul Schauble |
The wish list page contains these two items:
•Same, .NET
|
Copyright © 2026 University of California.
Permission is granted to copy, distribute and/or modify this document
under the terms of the GNU Free Documentation License,
Version 1.2 or any later version published by the Free Software Foundation.