Message boards : Number crunching : New work Discussion
Message board moderation
Previous · 1 . . . 46 · 47 · 48 · 49 · 50 · 51 · 52 . . . 91 · Next
Author | Message |
---|---|
Send message Joined: 15 May 09 Posts: 4541 Credit: 19,039,635 RAC: 18,944 |
I tried that, but it didn't help. :( Not even the deputy :) |
Send message Joined: 22 Feb 06 Posts: 491 Credit: 31,247,913 RAC: 15,591 |
Betelgeuse. What are your settings for days of work to be stored? The N216 models can easily take 15 to 20 days to complete and if BOINC thinks that you haven't allowed enough time it might not get you any work. Also are you running any other work units from other projects? |
Send message Joined: 9 Oct 20 Posts: 690 Credit: 4,391,754 RAC: 6,918 |
Betelgeuse.I'm not sure you're right there, I have all my machines set to 0+0.13 days = about 0+3 hours work. They had no trouble getting we a weather at home task (Windows) per core. |
Send message Joined: 6 Oct 06 Posts: 204 Credit: 7,608,986 RAC: 0 |
Betelgeuse.I'm not sure you're right there, I have all my machines set to 0+0.13 days = about 0+3 hours work. They had no trouble getting we a weather at home task (Windows) per core. ------------- Even if Alan is not correct, let the third party answer because Alan is talking common sense. Ten plus ten days of work in the settings fetches me twenty-four WU's. At least they used but I have a suspicion that the server settings have been tweaked a bit, got only twelve this time around. |
Send message Joined: 31 Aug 04 Posts: 10 Credit: 2,538,005 RAC: 0 |
Betelgeuse. OK, I received a task now.. My settings were store at least 0.5 days of work with an additional 0.1 days of work. Set all other projects (I am running 12 projects at the moment) to no new tasks and changed settings to 1.5 + 1.5 days. First try after this I got a task. A couple notes/questions related to this: - I enabled some of the event log debug options. This resulted in a huge number of lines in the event log, but nothing useful. - Why does CPDN not issue work when I request it ? The settings say store at least a certain amount of work, there is nothing about a maximum. - When I first encountered this problem I thought it was related to 32-bit library issues, but I have doubts about this now. I recently switched from CentOS 7 to Ubuntu 18.04 LTS because some of the projects I am running were dynamically linked to newer libraries than provided with CentOS 7. In these cases I would get tasks, but they would error out because of the missing dynamic libraries. Does the CPDN server has any clues as to the dynamic libraries installed on my system ? I would think not as there are too many different versions for different Linux distributions, so my guess is that it would just send out work and it would error out because of the missing dynamic libraries. Thanks everybody for the help. Tom |
Send message Joined: 9 Oct 20 Posts: 690 Credit: 4,391,754 RAC: 6,918 |
OK, I received a task now.. Perhaps in your case (and what happened with me) since you have work queued for other projects, it only wanted 1 CPDN task to give you enough buffer, that makes sense. I only got a few, until I suspended all other projects, then it filled the cores with 1 CPDN each, despite me having a 0+0.13 day buffer. But if you were getting no work from CPDN when you had a free core, something is up. Mine were initially just getting 1 task each time something else finished and a core was made free. But I definitely managed to fill a 24 core computer with 24 CPDN tasks when all other projects were suspended and the buffer was only 0+0.13 days. |
Send message Joined: 15 May 09 Posts: 4541 Credit: 19,039,635 RAC: 18,944 |
- When I first encountered this problem I thought it was related to 32-bit library issues, but I have doubts about this now. I recently switched from CentOS 7 to Ubuntu 18.04 LTS because some of the projects I am running were dynamically linked to newer libraries than provided with CentOS 7. In these cases I would get tasks, but they would error out because of the missing dynamic libraries. Does the CPDN server has any clues as to the dynamic libraries installed on my system ? I would think not as there are too many different versions for different Linux distributions, so my guess is that it would just send out work and it would error out because of the missing dynamic libraries. Don't know about CentOS but the instructions in the pinned post in the Linux section give instructions to install the libraries that are missing for most. If with any distro still getting problems with missing libraries, run ldd on the executable from CPDN and it will list anything missing. (I know from experience that it can then take a bit of detective work to find the package that supplies said library.) |
Send message Joined: 28 Jul 19 Posts: 150 Credit: 12,830,559 RAC: 228 |
Sadly this did not work for me. Having set nnt for all of my other projects I changed my settings from 0.1+0.1 days to 10+10 and it still did not bring any WUs down. Having reviewed the logs I sent in when I brought the up before the only thing I can see that’s odd is that the number of requested units appears to be zero even though the time requested is high. Any idea what controls this? |
Send message Joined: 9 Oct 20 Posts: 690 Credit: 4,391,754 RAC: 6,918 |
Sadly this did not work for me. I'm guessing, but could the zero be "not applicable" as is often the case in Boinc? Maybe it's supposed to request either a certain time or a certain number of units. What's in your logs with requests for other projects? |
Send message Joined: 31 Aug 04 Posts: 10 Credit: 2,538,005 RAC: 0 |
- When I first encountered this problem I thought it was related to 32-bit library issues, but I have doubts about this now. I recently switched from CentOS 7 to Ubuntu 18.04 LTS because some of the projects I am running were dynamically linked to newer libraries than provided with CentOS 7. In these cases I would get tasks, but they would error out because of the missing dynamic libraries. Does the CPDN server has any clues as to the dynamic libraries installed on my system ? I would think not as there are too many different versions for different Linux distributions, so my guess is that it would just send out work and it would error out because of the missing dynamic libraries. You are missing something here: Catch-22 If you get no tasks you also do not get the executable (you get it when the first task is assigned). It is very hard to run ldd on an executable you do not have. This is exactly the reason why I doubt the problem is caused by missing dynamic libraries: the CPDN server has no way of knowing if the correct libraries are present. If have seen this with several other projects: you get a task that errors out and in the error log you can usually see it was caused by missing dynamic libraries. You install the missing libraries (if they are provided by the distribution you are using !) and you check if the next task works. Tom |
Send message Joined: 9 Oct 20 Posts: 690 Credit: 4,391,754 RAC: 6,918 |
You are missing something here: Catch-22 Not meaning to be rude here, but.... Windows Boinc just works. I tried Linux once because someone said it was faster for Universe@home. Problem after problem just installing things. |
Send message Joined: 15 May 09 Posts: 4541 Credit: 19,039,635 RAC: 18,944 |
Not meaning to be rude here, but.... Windows Boinc just works. I tried Linux once because someone said it was faster for Universe@home. Problem after problem just installing things. However over the past year this project has had far longer without any tasks for Windows than for Linux. That may well change in the future as it has in the past which is why I also have the Windows version installed under WINE. The problem is not with BOINC but with the projects and their applications. I was told over on the BOINC fora that packing statically linked libraries with the science apps is unlikely to work because of incompatibilities with other libraries etc in distributions so even if it works on one distro it might not on another. Also many of the problems with BOINC itself on Linux are not down to BOINC but what the packagers do once they get hold of it. I have never had a problem with rolling my own on a clean install, only with incompatibilities after three or more upgrades further down the line from a clean install. If have seen this with several other projects: you get a task that errors out and in the error log you can usually see it was caused by missing dynamic libraries. You install the missing libraries (if they are provided by the distribution you are using !) and you check if the next task works. I really don't have a clue about the problem with getting no tasks downloaded but on a fresh install, I always start the work downloading, suspend computation and run ldd to check I have everything I need installed before starting the tasks. I know a bit of a faff especially for anyone used to Windows but still a lot easier getting things to work than it was when I first installed Linux in '99! |
Send message Joined: 9 Oct 20 Posts: 690 Credit: 4,391,754 RAC: 6,918 |
However over the past year this project has had far longer without any tasks for Windows than for Linux. That may well change in the future as it has in the past which is why I also have the Windows version installed under WINE.I was considering putting up with the horrid Linux interface on some machines I only use for Boinc to get more Boinc processing done, but it turns out only a small handful of apps in Boinc are faster in Linux, it just depends on who coded it. Most projects always have Windows work when they have work, so I'm happy with that. And when I looked at the recent posts of sets in another thread, it looks like last year had as much Windows as Linux work. The problem is not with BOINC but with the projects and their applications. I was told over on the BOINC fora that packing statically linked libraries with the science apps is unlikely to work because of incompatibilities with other libraries etc in distributions so even if it works on one distro it might not on another. Also many of the problems with BOINC itself on Linux are not down to BOINC but what the packagers do once they get hold of it. I have never had a problem with rolling my own on a clean install, only with incompatibilities after three or more upgrades further down the line from a clean install.Windows doesn't seem to have problems with libraries - it certainly used to, I remember 20-30 years ago having to download dll files for programs, which are I assume the equivalent. I really don't have a clue about the problem with getting no tasks downloaded but on a fresh install, I always start the work downloading, suspend computation and run ldd to check I have everything I need installed before starting the tasks. I know a bit of a faff especially for anyone used to Windows but still a lot easier getting things to work than it was when I first installed Linux in '99!At the risk of going off topic and having an admin yell at me, what advantages do you get running Linux? I used to be told "it's more stable", but I can't remember the last Windows crash. And I'd replace your "bit of a faff" with "a lot of a faff". All I did was try to install Boinc and two projects, then I start having to use a command line to give permission to install?! |
Send message Joined: 16 Jan 10 Posts: 1084 Credit: 7,884,997 RAC: 4,577 |
I was considering putting up with the horrid Linux interface on some machines I only use for Boinc to get more Boinc processing done, but it turns out only a small handful of apps in Boinc are faster in Linux, it just depends on who coded it. Most projects always have Windows work when they have work, so I'm happy with that. And when I looked at the recent posts of sets in another thread, it looks like last year had as much Windows as Linux work. Generally people choose Linux if it suits them for their own reasons or if work is only available on that platform. Microsoft has also annoyed people — including me — because of its update system, which can crash models without a bit of preventive registry editing, I am not aware of any consistent performance bias. Windows doesn't seem to have problems with libraries - it certainly used to, I remember 20-30 years ago having to download dll files for programs, which are I assume the equivalent. I’ve never had a BOINC library problem with Windows. At the risk of going off topic and having an admin yell at me, what advantages do you get running Linux? I used to be told "it's more stable", but I can't remember the last Windows crash. And I'd replace your "bit of a faff" with "a lot of a faff". All I did was try to install Boinc and two projects, then I start having to use a command line to give permission to install?! Microsoft has withdrawn its advice to upgrade to Windows 10 from earlier versions, and that certainly reflects my experience — I had to downgrade from Windows 10 to avoid random blue screens of death. Now I get warnings about updates being withheld but no Windows crashes. I don’t remember ever getting a system crash on Linux. In my experience the only reliable way to get the mods to yell at you is to obsess about credits and even then their patience is admirable. The mods and other interested users are in it for the science and here to help. |
Send message Joined: 9 Oct 20 Posts: 690 Credit: 4,391,754 RAC: 6,918 |
Generally people choose Linux if it suits them for their own reasonsIt seems outdated to me, you have to set everything up manually, all this command line stuff should have gone in this century. or if work is only available on that platform.In general (outside Boinc), surely Windows has more programs available? I'd hate to be a Linux or Mac user, go to download some freeware, then find there's only a Windows version. As for Boinc, CPDN is the only place I've found with a division. Most projects seem to have a Linux and Windows program that can take the same WU as input. Either that or they have so many work units that both are always available. Microsoft has also annoyed people — including me — because of its update system, which can crash models without a bit of preventive registry editing,Me too, VERY annoyed. And when I looked up how to bypass it (all my 6 systems now just sit with a quiet notification icon in the tray indefinitely), I find millions of people complaining about it, some who have lost money due to crashed work running overnight (why aren't Microsoft sued?!) How hard can it be for them to ask first? Or not close down if something is running! I once had it actually close a document I'd started working on and hadn't saved. Because it didn't have a name yet, it didn't save it, it just closed it. I am not aware of any consistent performance bias.Universe@home I think it was, one of their two apps runs 2-3 times faster under Linux than Windows (which I confirmed). But I couldn't find anything else with any significant change, so I stopped bothering. I’ve never had a BOINC library problem with Windows.Neither have I, but I have with other programs in Windows back in earlier Windows versions. Usually just freeware that doesn't include something and assumes you have it, or a precise version of it. The only recent one I had was a game that stored the dll in the wrong place. It was supposed to link with Steam or something and randomly didn't, causing the Steam dlls to be "missing". Microsoft has withdrawn its advice to upgrade to Windows 10 from earlier versions, and that certainly reflects my experience — I had to downgrade from Windows 10 to avoid random blue screens of death.I have never in my life seen a BSOD attributed to Windows. They're almost always hardware faults (mostly dodgy RAM), and very occasionally a dodgy driver, which Windows usually points out. Now I get warnings about updates being withheld but no Windows crashes. I don’t remember ever getting a system crash on Linux.Windows 95 used to run out of resources or whatever it called them, but that was about it. I don't know why people have stability problems and I don't! In my experience the only reliable way to get the mods to yell at you is to obsess about credits and even then their patience is admirable. The mods and other interested users are in it for the science and here to help.Oh I've been banned for using a signature, where apparently signatures aren't allowed in one particular thread, because the mod started the thread, uses a mobile phone to read the forum, and doesn't like the page getting too big! Actually I have a go at people on Rosetta about credits and the like. When Rosetta runs out of work, there are a number of users that start jumping up and down saying they should have been given warning so they can make plans. For crying out loud, the whole point of Boinc is it will switch to other projects automatically. I pointed out the science was the point of it, and they were probably too busy doing research to mollycoddle their users. One of them told me he was upset because it ruined his automated credit monitoring system he'd programmed. I may have got a little rude at that point. |
Send message Joined: 15 May 09 Posts: 4541 Credit: 19,039,635 RAC: 18,944 |
As for Boinc, CPDN is the only place I've found with a division. Most projects seem to have a Linux and Windows program that can take the same WU as input. Either that or they have so many work units that both are always available. CPDN used to have tasks that could run on any platform but I think from memory the WAH2 tasks had a problem on Linux that caused them all to crash and it was decided to make tasks single platform to decrease development times. A side effect is that there is some increase in the reliability of the science. When the OpenIFS tasks arrive on the main site whichever year that is, they will be Linux only as there isn't a Windows version of OpenIFS. (There is a Mac version but to date anyway there has been no testing of it on that platform for CPDN. It is 64bit so no issues with missing 32bit libraries there. They do need a lot of Ram and disk space however! |
Send message Joined: 5 Sep 04 Posts: 7629 Credit: 24,240,330 RAC: 0 |
Climate models only being available on one OS, is mostly to do with lack of staff and time. And some types of research are best suited to a particular model type. *************** A few young people run these models, so this board is meant to be kept child friendly. Anything otherwise will get a post hidden on the spot. |
Send message Joined: 9 Oct 20 Posts: 690 Credit: 4,391,754 RAC: 6,918 |
Climate models only being available on one OS, is mostly to do with lack of staff and time. I learned all the swearwords in primary school. Got me quite a few detentions, especially from the religious nut teacher. It's illogical for adults to hide words from kids that already know them. |
Send message Joined: 28 Jul 19 Posts: 150 Credit: 12,830,559 RAC: 228 |
Sadly this did not work for me. I've repeated the attempt to get a new set of logs to work on. I set NNT for all other projects, set work request debug on, changed the buffer size from 0.1 + 0.1 to 10 + 10. Clomate prediction then refused to request work because it was backed off????? (no backoff showing in manager and I'd done nothing on it for days). so I removed the project and reattacted producing the following log :- Wed 11 Nov 2020 12:43:36 GMT | climateprediction.net | Resetting project Wed 11 Nov 2020 12:43:36 GMT | climateprediction.net | Detaching from project Wed 11 Nov 2020 12:43:36 GMT | | [work_fetch] Request work fetch: Detach Wed 11 Nov 2020 12:43:36 GMT | | [work_fetch] Request work fetch: project detached by user Wed 11 Nov 2020 12:43:40 GMT | | choose_project(): 1605098620.220595 Wed 11 Nov 2020 12:43:40 GMT | | [work_fetch] ------- start work fetch state ------- Wed 11 Nov 2020 12:43:40 GMT | | [work_fetch] target work buffer: 864000.00 + 864000.00 sec Wed 11 Nov 2020 12:43:40 GMT | | [work_fetch] --- project states --- Wed 11 Nov 2020 12:43:40 GMT | | [work_fetch] --- state for CPU --- Wed 11 Nov 2020 12:43:40 GMT | | [work_fetch] shortfall 20552559.01 nidle 0.00 saturated 9102.83 busy 8831.65 Wed 11 Nov 2020 12:43:40 GMT | | [work_fetch] ------- end work fetch state ------- Wed 11 Nov 2020 12:43:40 GMT | | [work_fetch] No project chosen for work fetch Wed 11 Nov 2020 12:44:28 GMT | | Fetching configuration file from https://climateprediction.net/get_project_config.php Wed 11 Nov 2020 12:44:40 GMT | | choose_project(): 1605098680.342802 Wed 11 Nov 2020 12:44:40 GMT | | [work_fetch] ------- start work fetch state ------- Wed 11 Nov 2020 12:44:40 GMT | | [work_fetch] target work buffer: 864000.00 + 864000.00 sec Wed 11 Nov 2020 12:44:40 GMT | | [work_fetch] --- project states --- Wed 11 Nov 2020 12:44:40 GMT | | [work_fetch] --- state for CPU --- Wed 11 Nov 2020 12:44:40 GMT | | [work_fetch] shortfall 20553279.97 nidle 0.00 saturated 9025.78 busy 8771.65 Wed 11 Nov 2020 12:44:40 GMT | | [work_fetch] ------- end work fetch state ------- Wed 11 Nov 2020 12:44:40 GMT | | [work_fetch] No project chosen for work fetch Wed 11 Nov 2020 12:45:00 GMT | climateprediction.net | Fetching scheduler list Wed 11 Nov 2020 12:45:01 GMT | climateprediction.net | Master file download succeeded Wed 11 Nov 2020 12:45:01 GMT | | [work_fetch] Request work fetch: Master fetch complete Wed 11 Nov 2020 12:45:06 GMT | climateprediction.net | piggyback_work_request() Wed 11 Nov 2020 12:45:06 GMT | | [work_fetch] ------- start work fetch state ------- Wed 11 Nov 2020 12:45:06 GMT | | [work_fetch] target work buffer: 864000.00 + 864000.00 sec Wed 11 Nov 2020 12:45:06 GMT | | [work_fetch] --- project states --- Wed 11 Nov 2020 12:45:06 GMT | climateprediction.net | [work_fetch] REC 0.000 prio -0.000 can request work Wed 11 Nov 2020 12:45:06 GMT | | [work_fetch] --- state for CPU --- Wed 11 Nov 2020 12:45:06 GMT | | [work_fetch] shortfall 20553504.36 nidle 0.00 saturated 9006.52 busy 8741.00 Wed 11 Nov 2020 12:45:06 GMT | climateprediction.net | [work_fetch] share 1.000 Wed 11 Nov 2020 12:45:06 GMT | | [work_fetch] ------- end work fetch state ------- Wed 11 Nov 2020 12:45:06 GMT | climateprediction.net | piggyback: resource CPU Wed 11 Nov 2020 12:45:06 GMT | climateprediction.net | [work_fetch] set_request() for CPU: ninst 12 nused_total 0.00 nidle_now 0.00 fetch share 1.00 req_inst 12.00 req_secs 20553504.36 Wed 11 Nov 2020 12:45:06 GMT | climateprediction.net | [work_fetch] request: CPU (1.00 sec, 0.00 inst) Wed 11 Nov 2020 12:45:06 GMT | climateprediction.net | Sending scheduler request: Project initialization. Wed 11 Nov 2020 12:45:06 GMT | climateprediction.net | Requesting new tasks for CPU Wed 11 Nov 2020 12:45:07 GMT | climateprediction.net | Scheduler request completed: got 0 new tasks Wed 11 Nov 2020 12:45:07 GMT | climateprediction.net | No tasks sent Wed 11 Nov 2020 12:45:07 GMT | climateprediction.net | Project requested delay of 3636 seconds Wed 11 Nov 2020 12:45:07 GMT | | [work_fetch] Request work fetch: RPC complete |
Send message Joined: 15 May 09 Posts: 4541 Credit: 19,039,635 RAC: 18,944 |
I have a full buffer at the moment but when it gets low I will enable the flag so I can compare what I have on a system that does get work. I am struggling to understand this one. |
©2024 cpdn.org