Posts by David Ball
|
21)
Message boards :
Projects :
News on Project Outages
(Message 52923)
Posted 3 Mar 2014 by David Ball Post: I just checked again and Asteroids@home is downloading work OK now. |
|
22)
Message boards :
Projects :
News on Project Outages
(Message 52908)
Posted 2 Mar 2014 by David Ball Post: Asteroids@home is getting download failures on all new work. You can reach the website and change it's preferences and the client will pick it up. You can request jobs and get them but they will always fail with a download error. |
|
23)
Message boards :
Questions and problems :
1,000,000 WORK UNITS!!!!! now how do I sign on to Einstein?
(Message 52438)
Posted 8 Feb 2014 by David Ball Post: Thanks for the correction. For some reason, I had never seen that page, just observed the extended log info and read about it in various forums. Actually, on my dedicated crunchers that do have GPUs, I tend to set CPU/GPU projects to GPU only and then assign then resource shares that add up to 1.00 so that I can schedule GPU usage separate from CPU projects. For instance, I might have: project resource share SETI 0.34 Einstein 0.33 GPUGRID 0.33 ---- Total 1.00 Then I set my CPU only projects so that they total 99.00 which basically lets me set the percentage I want them to use as the resource share. Unfortunately, I don't recall how I had WCG setup when it had several CPU only subprojects and 1 GPU capable project. I had been hoping that BOINC would eventually assign separate CPU and GPU resource shares but after reading that document page I can see why they have it like it is. Thanks! |
|
24)
Message boards :
Questions and problems :
1,000,000 WORK UNITS!!!!! now how do I sign on to Einstein?
(Message 52313)
Posted 3 Feb 2014 by David Ball Post:
IIUC, The BOINC client tries to keep projects balanced based on how much credit they've earned in the 10 days prior to the current time. When you add a new project, it tends to get a lot of work the first few days until it catches up with the existing projects.I believe this is called the creditnew system. Since SETI awards lousy credits for GPU work, I believe there is some workaround that other projects have developed to trick the BOINC client into thinking they award the same credits as SETI so that the percentages become time based rather than credit based. In this case, Einstein will run a lot until it catches up with SETI on the amount of time spent on it in the 10 days prior to the current time. FYI, the BOINC client does keep track of when work units are due and will give them priority on running if it thinks that they might miss their deadlines. Those SETI work units that you already have sitting on your machine waiting to run should run in time to meet their deadline. I use some debug options to get the BOINC client to log a bunch of work fetch stuff. The following is how it calculates the priority for work fetch. Please note that this machine has some projects suspended and some configured not to run the CPU or GPU on this machine because this is the machine where I monitor the totals from my dedicated crunchers. Just ignore anything with a prio of zero since those projects are either suspended or haven't had any work available in the last 10 days. I believe that REC stands for recent credit. The prio is calculated using the recent credit and project percentage. I'm not sure but I think the projects that want to be time based rather than credit based lie to the BOINC client about the recent credit so that it will match the recent credit that SETI would have gotten if it had run the same number of hours as that project did.
Basically, this is the reason your machine is running a lot of Einstein@Home currently. When Einstein was added, it's REC was zero. When Einstein catches up on the amount of time spent on it during the previous 10 days, it should run 50/50 as you've selected. As long as Einstein has work available each time it's asked for it, you might see it run Einstein for 5 days before it starts fetching Einstein and Seti work units equally. If a project is asked for work and doesn't return anything, it will ask for work from the next lower priority project. Since projects have to do a lot of maintenance on their servers and occasionally move to new servers or have server failures, don't expect it to ever be perfectly even. For instance, malariacontrol is usually extremely reliable but it has some kind of problem and has been down for almost 10 days. Many of my machines have work they fetched before malariacontrol went down and are still trying to upload or report that work. Some of the smaller projects like correlizer, Volpex, or fightmalaria may go several weeks without having work available. Hope this helps. |
|
25)
Message boards :
Projects :
News on Project Outages
(Message 52232)
Posted 31 Jan 2014 by David Ball Post: Volpex ( http://volpex.cs.uh.edu/VCP/ ) hasn't had work for a while and now the website has been down for about 4 days. Does anyone know if the project is down permanently? |
|
26)
Message boards :
Projects :
News on Project Outages
(Message 52212)
Posted 29 Jan 2014 by David Ball Post: Any news on when Malariacontrol will be back online? All these WU trying to upload or waiting to report are a pain, especially as they are now overdue. Since projects that have big crashes often have to restore from a backup and lose the record of outstanding WU, I'm wondering whether to reset the project to try to get rid of them but I'm not exactly sure what a project reset from BOINC manager would do if the project is unreachable. malariacontrol.net went offline at around 0900 UTC on Friday 24 January and will be down for at least this weekend. All pages are currently generating the following message: |
|
27)
Message boards :
Projects :
News on Project Outages
(Message 51198)
Posted 7 Nov 2013 by David Ball Post: There is a big problem developing with MindModeling@Home (Beta) !!! They are apparently trying to switch to a new machine for at least part of the project and are starting to use links to the url http://192.168.108.111 for portions of the website. The instructions for joining the project are also saying Join MindModeling@Home (Beta) If you are already a member, you get the following in your project responses 11/7/2013 11:17:32 AM | MindModeling@Beta_WSU | You used the wrong URL for this project. When convenient, remove this project, then add http://192.168.108.111/ Also, downloads are starting to fail (probably uploads will also fail). This url will NOT work over the internet. The subnet 192.168.x.x is reserved as a private address for local networks. You can find info on this in rfc1918 (Address Allocation for Private Internets) at http://tools.ietf.org/html/rfc1918 The Internet Assigned Numbers Authority (IANA) has reserved the following three blocks of the IP address space for private internets: 10.0.0.0 - 10.255.255.255 (10/8 prefix) 172.16.0.0 - 172.31.255.255 (172.16/12 prefix) 192.168.0.0 - 192.168.255.255 (192.168/16 prefix) They need to setup the change at their router. edit: I've tried to contact them but I don't know if they will see the message because I used the old url since the new one won't work. |
|
28)
Message boards :
Projects :
QMC News - Project Moving, URL Change
(Message 50817)
Posted 9 Oct 2013 by David Ball Post: I'm having the same problem. I have a 42 hour WU that it's been trying to turn in but gets the feeder off-line error. The WU will be past due in about 1.5 hours. Does anyone know if QMC will accept past due WU if it ever comes back up? I was able to go into my account and reduce the resource share for QMC but the feeder off-line error also prevents the client from getting the new resource share. |
|
29)
Message boards :
Questions and problems :
10 projects and no new tasks for weeks..
(Message 50634)
Posted 25 Sep 2013 by David Ball Post: I added right now a new project SAT@home and superlinkattechnion, the first one is downloading packages at the moment, I'm going to remove all the projects and re-add them again, let's see what happen. I don't know if you're familiar with superlinkattechnion but, FYI, they often go very many months without having work for users so don't be surprised if you don't get any work from them. -- David |
|
30)
Message boards :
Projects :
News on Project Outages
(Message 50633)
Posted 25 Sep 2013 by David Ball Post: Asteroids@Home update According the most recent post, early on Sep 23, on their twitter account (https://twitter.com/AsteroidsAtHome) the disks were incompatible between the old and new machines so they're transferring the data via RSYNC and making some changes so that the new server can run BOINC as soon as possible. Currently (Late Tuesday night in the eastern US), The apache server is running but it just says Not Found -- David |
|
31)
Message boards :
Projects :
News on Project Outages
(Message 49776)
Posted 3 Jul 2013 by David Ball Post: Have been unable to reach Ibercivis to turn in work for a few days. 7/3/2013 9:07:51 AM | ibercivis | Sending scheduler request: Requested by user. The completed task is still sitting in the tasks list on the boinc client. It was received on June 20th. It finished late Friday or over the weekend, IIRC. I remember thinking on Monday that, since it was a weekday, maybe they would fix the problem and it would finally go away. |
|
32)
Message boards :
Projects :
News on Project Outages
(Message 49231)
Posted 21 May 2013 by David Ball Post: I've been unable to get any work from mindmodeling@home for a few days and for the last day or so, I get a completely blank web page when looking at the MindModeling home page. isup.me says: It's not just you! http://mindmodeling.org looks down from here. |
|
33)
Message boards :
Projects :
LHC sixtrack - no work?
(Message 49230)
Posted 21 May 2013 by David Ball Post: I've been getting new work from LHC but it just seems to be available randomly during the day. I'm only running a small share of LHC so it's not a problem for me but if you're trying to get a lot of workunits, it might be a problem. Also, boinc 7.0.64 seems to increase the time it will wait before it tries the project again each time it fails to get work. If you will occasionally, do a manual update on LHC, it will reduce the wait time. Note also that if it doesn't get work from LHC and it can get work from another project, it will. They're working on changing the work fetch logic and it will probably be backported to the 7.0.x version once it's debuged. Do NOT run the 7.1.x client because they've just started on it and there are lots of work fetch problems in it. Boinc 7.1.1 is listed as the UNSTABLE alpha test version and it really is UNSTABLE :-) David |
|
34)
Message boards :
Questions and problems :
BOINC starving while waiting for memory
(Message 49152)
Posted 16 May 2013 by David Ball Post: When I looked at your RNA WU, it says that estimated runtime on reference system 11w 2d 22h 12m 14s (6905534.7950536 s) That's 11 weeks of runtime and RNA doesn't checkpoint!!! Read the following thread in the RNA message boards: checkpoints for long WU The recommended method of running these RNA WUs seems to be to run a Virtualbox virtual machine (20 GB virtual HDD and 1 GB memory + 2 GB memory per cpu core on the virtual machine so for a dual core machine you're talking 5 GB memory allocated for the VM). The virtual machine will need to run on a 64 bit OS. Then you snapshot the virtual machine regularly and restart it from the latest virtual machine snapshot each time you restart the physical machine. David |
|
35)
Message boards :
Projects :
Resources wasted on Primegrid project?
(Message 47697)
Posted 9 Feb 2013 by David Ball Post: How many prime numbers or ET will you find if you're dead? None. That's why investigating on health is more important. First things first. I'd prefer to run health projects so I run POEM (when they're not out of WUs) with Einstein as backup and primegrid as a secondary backup if neither of those has GPU work. Unfortunately, WCG wont run on my old nVidia 9600 GT, probably because it doesn't have double precision floating point. WCG doesn't give a reason, it just says "The Help Conquer Cancer NVIDIA GPU application cannot run on your graphics card". I'm not a gamer so I just have a low end graphics card. I'm going to update to a more modern card that has DP floating point but it will still be a low end one. That will also get me more modern HD video decoders on the card and cut down on power usage. David |
|
36)
Message boards :
Questions and problems :
Bug in newer server code for computer preferences ?????
(Message 47333)
Posted 17 Jan 2013 by David Ball Post: Machine info: Multiple q6600 quads running Vista 32 and BOINC 7.0.28 with "home" compute preferences. One C2D E6420 (basically conroe 6400 but with all 4 megs of L2 cache) running Vista 32 and BOINC 7.0.28 with "work" compute preferences. I normally maintain my "Computing Preferences" in my account on the SIMAP website. I keep one set of "default" preferences which runs on the quads and a set of "work" preferences which run on the C2D which has a nVidia 9600 GT graphics card. Recently, I've joined some new projects and I happened to edit my "computing preferences on one of those projects (fightingmalaria@home IIRC). I just made a simple change to one of the (default IIRC) preferences like time slice or how much work to fetch. It appeared to work so I didn't really look through the 2 sets of preferences in detail on the project website. Once I had gone through my 3 crunchers and had them contact that project to get the updated "computing preferences", I noticed that something seemed strange. It appears that the project had randomly combined values from the default and work preferences and had sent "computing preferences" to my machines that only had the merged "default" computing preferences and no "work" computing preferences. They had already been propagated to other projects so I ended up having to go back to SIMAP, fix the merged default settings and re-create the work settings. Again, only the "Computing preferences" were scrambled, afaik. I wrote it off to a one time glitch but it has happened a couple of more times, once on fightingmalaria and once on another recently joined project. I don't recall what the other project was but it was either a new project or LHC, which I re-joined now that they have work again. As long as I change the preferences at SIMAP or Seti (when Seti is up), I never have a problem. Here's the image with my stats so you can see what projects I am attached to.  David Ball |
|
37)
Message boards :
Projects :
Hydrogen@home
(Message 36589)
Posted 26 Jan 2011 by David Ball Post: Will the GPU apps require double precision floating point or will they work on lower end graphics cards like nVidia 9600 GT or ATI HD 5770 and below (58xx are the only 5000 series with DP math IIUC).? |
|
38)
Message boards :
Projects :
QuantumFIRE
(Message 36470)
Posted 18 Jan 2011 by David Ball Post: Do you know if QuantumFIRE will have the same URL when it returns? I'm wondering if there's any point to leaving it attached at the current URL in my BOINC client. |
|
39)
Message boards :
Projects :
News on Project Outages
(Message 36355)
Posted 6 Jan 2011 by David Ball Post: The Rosetta web server now says The project's fileserver has crashed. We're working to get things back online as soon as possible. Thanks for your patience. -KEL 01/06/2011 |
|
40)
Message boards :
Projects :
News on Project Outages
(Message 34257)
Posted 17 Aug 2010 by David Ball Post: Anybody know what happened with Virtual Prairie? They've been down completely for many days. |
Previous 20 · Next 20
Copyright © 2024 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.