Info | Message |
---|---|
1) Message boards : Questions and problems : "Communication deferred 24 hours ..."
Message 116153 Posted 27 days ago by Erich56 |
Problem solved: by coincidence, I noticed that the IP address of the notebook was changed (no idea how come, because on the router I have set it as fixed). And for this reason, the squid proxy did not allow the connection to LHC. So I added the new IP address to the hosts list in the squid.config of the proxy, and now it works :-) |
2) Message boards : Questions and problems : "Communication deferred 24 hours ..."
Message 116151 Posted 27 days ago by Erich56 |
So, what I did now was: The counter showed 21:58 hours, I closed the Boinc Manager and the Boinc client and ran the Boinc "repair" installation. Even this did now help, the counter jumped back to 24 hours. It's a conondrum :-( |
3) Message boards : Questions and problems : "Communication deferred 24 hours ..."
Message 116149 Posted 27 days ago by Erich56 |
In reply to Dave's message of 23 May 2025:I think this is an LHC issue and that when tasks are aborted, they impose a 24 hour back off till you can get more tasks and there is nothing you can do but wait. Have you tried the LHC forums? There should be someone there who can answer this.Yes, someone very knowledgeable from there gave me the advice to restart the Boinc client. As mentioned above, this did not help though. Further, I don't think the problem is related to LHC specifically. There have been many cases that I aborted tasks, over the years (for one reason or the other), but never ever the Boinc Manager bahaved the way it does now. Also, if there was indeed this 24 hour limit, why does the counter still jump back to 24 hours each time I push the "update" button? I am sure that the problem is related to BOINC. And I am even not sure what will happen after the 24 hours are gone. Maybe it then again jumps back to 24 hours the moment I push "update". Well, I can wait till tomorrow, no problem. If the situation then is still the same, I guess all I can do is to uninstall and reinstall BOINC. |
4) Message boards : Questions and problems : "Communication deferred 24 hours ..."
Message 116146 Posted 27 days ago by Erich56 |
After several LHC/CMS tasks had crashed, they were shown in the BOINC Manager with "computation error", and I was not able to remove these tasks from the Manager (however, I could remove them in the Oracle Virtual Box Manager). As long as this was the case, no new tasks could be downloaded; no way to get these entries removed from the BOINC Manager. So I decided to reset the LHC project, which was successful in a way that the crashed tasks did no longer show up in the BOINC Manager. However, if I click on "update" for getting new tasks, the BOINC Manager in the "status" column shows "communication deferred 24 hours, and this is now counting downwards every minute, for the next 24 hours :-( In other words, I will have to wait for a full day in order to get new tasks downloaded. What I also tried about 1 hour later was to restart the BOINC client (as well as the BOINC Manager) - and while before doing this, the remaining time was shown as some 23 hours, after restart it had jumped back to 24 hours and is now again counting down from there. What's going wrong? Any ideas how to solve the problem ? |
5) Message boards : GPUs : BOINC does not recognize the correct GPU driver
Message 115943 Posted 20 Apr 2025 by Erich56 |
In reply to Grant (SSSF)'s message of 20 Apr 2025:Failed to delete old coproc-info.xml. error code - 110.That is what happens if you make the file read only, as per my question earlier in the thread. well, I remember your posting: "You haven't previously set any BOINC configuration files to read only to stop them from changing?" - in fact, I had never set any BOINC configuration files to "read only". The issue did not occur after having updated the video driver. I hadn't done anything, as I explained anyway. The problem came up out of the blue, without any changes from my side. But I am not worried too much now, because I am sure with the latest driver now, for long time GPUGRID (and other projects) will be able to function. And even if I upgrade the driver one day, there will not be a problem as long as the driver shown in the coproc-info.xml works with the given task. Plus, I now know anyway how to solve the problem, should one ever come up :-) BTW: the host just happened to get a GPUGRID task, and it runs well (although it may fail after some time with the "energy is NAN" error like so many GPUGRID tasks in the past - but this will then have nothing to do with my recent driver problem). |
6) Message boards : GPUs : BOINC does not recognize the correct GPU driver
Message 115941 Posted 20 Apr 2025 by Erich56 |
I now downloaded the latest version of DDU - 18.1.1.0 (before I used 18.1.0.0) and ran the tool three times - for AMD, Intel and NVIDIA. In save mode, of course, and network cable unplugged. Then, before re-installing the NVIDIA driver 572.83, I started BOINC, opened the event log, and again the driver 516.94 was shown - for CUDA as well as for OpenCL. And: one line above it says: Failed to delete old coproc-info.xml. error code - 110. So I went on to install driver 572.83, and - as expected - the BOINC event log again shows driver 516.94. Then I opened the BOINC folder and the coproc-info.xml (dated some time in 2022). On various places, the driver was shown as 51694 or 516.94. So I changed these entries to read 57283 or 572.83, saved the file and restarted BOINC. And, indeed, now the event log shows 572.83, and when I tried to download a GPUGRID task (however, there wasn't any available), I no longer got the request to update the GPU driver. Unbelievable, how simply the problem finally got solved :-) |
7) Message boards : GPUs : BOINC does not recognize the correct GPU driver
Message 115924 Posted 17 Apr 2025 by Erich56 |
In reply to Jord's message of 16 Apr 2025:Did you run DDU only for Nvidia drivers?yes, I ran DDU for Nvidia only. So thanks for the hint, I will run DDU again choosing NVIDIA, AMD and Intel. Still it is a conondrum to me what happened totally out of the blue. Without having done anything from my side, BOINC all of a sudden decided to no longer recognize the at that time current driver 537.99, but insists on driver 516.94. While all other tools like the NVIDIA Inspector, GPU-Z, and the NVIDIA System Control showed 537.99. I also started Folding@Home - and even there 537.99 was shown. And all the before said is true after having had installed the latest driver 572.53 - this version is shown everywhere, except in BOINC. So in my opinion the problem clearly is with BOINC, right? |
8) Message boards : GPUs : BOINC does not recognize the correct GPU driver
Message 115923 Posted 17 Apr 2025 by Erich56 |
In reply to Grant (SSSF)'s message of 16 Apr 2025:You haven't previously set any BOINC configuration files to read only to stop them from changing? (something i had to do some time back due to issues with the video driver (an Intel OpenCL entry in the registry broke the support for NVidia OpenCL & CUDA. No OpenCL, but 2 CUDA devices for just one video card...)).Not that I can remember |
9) Message boards : GPUs : BOINC does not recognize the correct GPU driver
Message 115910 Posted 15 Apr 2025 by Erich56 |
In reply to Ian&Steve C.'s message of 15 Apr 2025:when you ran DDU, did you go into the settings to prevent Windows from downloading drivers automatically through Windows Updates?during applying DDU and installing the latest driver afterwards, the network cable was unplugged. if DDU isnt removing it for some reason. I would try installing the OS freshI was afraid that this would be the only solution, after nothing else helps :-( |
10) Message boards : GPUs : BOINC does not recognize the correct GPU driver
Message 115905 Posted 15 Apr 2025 by Erich56 |
In reply to Jord's message of 15 Apr 2025:My advice is to run Display Driver Uninstaller from https://www.guru3d.com/download/display-driver-uninstaller-download/ in Windows Safe Mode and after cleaning everything and rebooting, to reinstall the newest drivers and try again. as said in my posting - i did run DDU (forgot to mention that I followed the recommendation to run it in safe mode) before installing the latest driver. the very strange thing is, as mentioned: until about 2 weeks ago, everything worked fine. An from one moment to the other (I did not make any changes either to hardware or to software) BOINC showed driver 516.94. Too bad that GPUGRID, one of my favorite projects, does not run with this alleged old driver. Other projects do. |
11) Message boards : GPUs : BOINC does not recognize the correct GPU driver
Message 115902 Posted 15 Apr 2025 by Erich56 |
Until about 2 weeks ago, I have been crunching GPU tasks from various projects on my NVIDIA Quadro P5000 without any problem, mostly GPUGRID whenever tasks were available. The driver version was 537.99. Then, all of a sudden when trying to download further GPUGRID tasks, the BOINC event log told me to update the GPU driver. So, first I restarted BOINC and noticed in the event log that the driver was shown as version 516.94 (!!!). No idea how come - maybe this was the driver version at time of installation the GPU years ago. And of course, when trying to download GPUGRID tasks, again I was requested to update the driver. So, obviously this driver version is too old for GPUGRID (although PrimeGrid and MooWrapper GPU tasks worked well). Okay, next I downloaded the newest driver from the NVIDIA site, version 572.83, and installed it after having removed the former driver (537.99) via DDU. Everything worked well, and the NVIDIA Inspector, GPU-Z as well as the NVIDIA Control Panel now showed version 572.83. Then I started BOINC, and annoying enough, at the startup phase the event log still complains about the outdated driver version 516.94, and hence would not let me download GPUGRID tasks (regardless whether there are any available or not). So I have no idea at this point what I can do in order to make BOINC (version is 8.0.2) recognize the driver 572.83. |
12) Message boards : Questions and problems : Can the "slots" folder in the Data directory be located on a seperate drive?
Message 112575 Posted 23 Aug 2023 by Erich56 |
I know that the complete BOINC Data directory (containing, among others, die "slots" folder) can be transferred to a different drive by modifying the "DATADIR" entry in the Windows registry. Can anyone tell me whether there is a way to put the "slots" folder only onto a different drive (e.g. Ramdisk)? |
13) Message boards : Questions and problems : "job cache full" due to indicated unrealistic high remaining runtimes of GPUGRID Python tasks
Message 109980 Posted 4 Oct 2022 by Erich56 |
For about a week, I have the following problem when trying to crunch GPUGRID Python tasks on one of my hosts which consists of: 2 CPUs Xeon E5 8-core / 16-HT each. 1 GPU Quadro P5000 128 GB Ramdisk 128 GB system memory until about 10 days ago, I ran 2 Pythons simultaneously (with a setting in the app_config.xml: 0.5 gpu usage) without any problems. Now, while only 1 Python is running and I push the update button on the BOINC manager for fetching another Python, the BOINC event log tells me that no Pythons are available. Which is not the case though, as the server status page shows plenty unsent tasks for download; besides, I can download such tasks on another PC. So I tried to download tasks from other GPU projects, and in all cases the event log says: not requesting tasks: don't need (CPU; NVIDIA GPU: job cache full). In the BOINC computing preferences, I then set the "store at least work" to the maximum possible of 10 days, and under "store up to an additional" also 10 days. However, this did not solve the problem. There is about 94GB free space on the Ramdisk, and some 150GB free system RAM. So, what I noticed then was that "job cache full" is obviously preventing more downloads: a running Python, due to it's technical nature, shows remaing runtimes of 30 days, 60 days or even more - and this causes "job cache full" :-( Which does not reflect the reality though, because a Python, on this host, finishes after about 24 hours. Before, like on my other machines, remaining runtime for Pythons was indicated as 1-2 days. So, at some point something unknown must have happened, letting the runtimes jump up that drastically only on this one of my several hosts (in fact, another host whith a smaller GPU, right now, after a runtime of several hours, shows a remaining runtime of about 24 hours). I posted this problem in the GPUGRID forum already; members confirmed that due to the technical nature of the Python tasks, such unrealistic "remaining time" indications are shown in the BOINC manager; so, obviously in some cases that's normal what concerns the GPUGRID Python tasks. One member was even talking about 157 remaining days (!) shown on his host (while the tasks in fact finish far below 24 hours). With the generous hardware ressources of this host, I would like to crunch 2 Python tasks simulteaneously in any case (like I am doing it on another host with less hardware ressources). Can anyone help me to get out of this problem? Is there any possiblity to tweak the values or whatever? In the GPUGRID forum, so far I could not get any advice. |
14) Message boards : Questions and problems : BOINC not working with data directory on USB 3.0 disk ?
Message 104227 Posted 1 May 2021 by Erich56 |
Jord, it's exactly as you say - there are quite a number of different subtypes of USB3 ports. I first connected the external drive, which is USB3.0, to these ports at the rear of the case, i.e. to the motherboard directly. This did not work, though - I couldn' even get the curser moved. So I connected it to a port on fhe front of the case, called "USB3.0". The downside is, that this port is not part of the motherboard, but rather connected to it via internal cable. So my guess is that for this reason the connection simply is not stable enough for BOINC. What I decided this morning was to install Primo RAMDisk, with which I had working already on another PC several years ago. This new PC has 64GB RAM inside, so it's perfect for RAMDisk. An in fact, it works excellently :-) |
15) Message boards : Questions and problems : BOINC not working with data directory on USB 3.0 disk ?
Message 104214 Posted 30 Apr 2021 by Erich56 |
Depending on operating system, make sure the USB ports/controller don't go into power saving mode after N minutes.hm, I was even not aware that there is a setting for this. The OS is Windows 10 Pro. Where do I find this setting? I see the USB Controller in the Device Manager, but no such settings. |
16) Message boards : Questions and problems : BOINC not working with data directory on USB 3.0 disk ?
Message 104211 Posted 30 Apr 2021 by Erich56 |
In order not to overtax my SSD with the current stress test project from WordCommunityGrid, I decided to do what I had done already a few years ago when I participated in a project with enormous amounts of data written to the SSD: I shifted BOINC over to an external USB3-disk. And that worked out very well (although I had to accept the lower speed, of course) However, this time it did not seem to work. The BOINC manager gets disconnected every few minutes and needs to get reconnected manually. So I must consider this approach as failed, as sorry as I am for my SSD. Or does anyone here have any idea what's going wrong and what I could do to get the thing run properly? |
17) Message boards : Questions and problems : Error 2718. Missing package name for product code '{B3B5C017-61CD-4587-9FA8-7E5E2D94810E}
Message 103768 Posted 31 Mar 2021 by Erich56 |
...Richard, you made my day :-) The upgrade to 7.16.11 worked perfect! Thanks you so much for your quick help, best regards |
18) Message boards : Questions and problems : Error 2718. Missing package name for product code '{B3B5C017-61CD-4587-9FA8-7E5E2D94810E}
Message 103765 Posted 31 Mar 2021 by Erich56 |
On one of my PCs, long time ago I had installed BOINC 7.6.22 this way: Program Directory in Drive C:\ (SSD) and Data Directory in Drive D:\ (HDD). Now the HDD gave up. So I decided to download and install the latest version 7.16.11, completely on Drive C:\ As expected, I could not simply install the new version over the old one. So I tried to run the uninstall routine, which did not work either (obviously simply because there was no access to drive D:\ any more). So I deleted some BOINC-related files in the windows explorer of Drive C:\ manually, but many of them could not be deleted. So I opened the registry and deleted everything related to BOINC. After restarting the PC and beginning with the installation of 7.16.11, I got an error notice saying Error 2718. Missing package name for product code '{B3B5C017-61CD-4587-9FA8-7E5E2D94810E} As this looked like a registry entry I suspected that while deleting BOINC-related data from the registry before, I inadvertantly deleted something else which I should not have deleted. So I restored the registry and tried again to install 7.16.11. However, the same error message is showing up again. So obviously this does not have anything to do with the registry. However, I now have no idea how I can install BOINC. What does "Missing package" mean in this context? Hope that anyone here can help, thanks in advance[/url] |
19) Message boards : Questions and problems : can the "TdrLevel" check be switched off?
Message 87513 Posted 8 Aug 2018 by Erich56 |
In a forum someone pointed out that this may have to do with the TDR (Timeout Detection and Recovery), part of the WDDM which is part of Windows Vista and later, but NOT XP. Hence, the advice I got was to "switch off" this feature in Windows10 by adding the following registry entry in HKEY_LOCAL_MACHINE\SYSTEM\ControlSet001\Control\GraphicsDrivers: QWord "TdrLevel" with value "0". This, so to say, imitates what I had in WindowsXP, i.e. NO Tdr at all. Edit: I now tried this registry entry as DWORD (vice QWORD as before), and after this, BOINC did NOT complain about any "unsuable GPU". So I am wondering whether DWORD or QWORD is correct. I have been told that 64-bit systems use QWORD. However, all the other entries I saw in this specific section of the registry were DWORD. Maybe the reason why BOINC doesn't complain about this DWORD entry is that it doesn't have any effect. No idea. |
20) Message boards : Questions and problems : can the "TdrLevel" check be switched off?
Message 87509 Posted 8 Aug 2018 by Erich56 |
Boinc version 7.12.1 (x64), OS: Windows10 Home. 2 GTX980ti. Intel i7-4930k, running @ 3.9GHz On a WinXP machine with which I have been crunching various BOINC projects for the past 2 1/2 years without any problems, I now installed Windows 10. When crunching GPU projects (any one) on Windows 10, it happens in irregular intervals (maybe after 1 hour, or after 8 hours, or at some time inbetween) that all of a sudden the monitor freezes, and crunching stopps. Neither keyboard nor mouse react any more, so all I can do is pushing the off-button on the PC and make a reboot. Then, the Windows event log under "system" shows the warning "the graphic driver nvlddmkm does no longer react and was restored". This notice shows up every 4 seconds, from the time on the crash happened until I switched off the PC. So if this happens during night and I notice it only next morning, this entry is logged a few thousand times. Under "details" it shows "eventID 4101", and under event data "nvlddmkm". In a forum someone pointed out that this may have to do with the TDR (Timeout Detection and Recovery), part of the WDDM which is part of Windows Vista and later, but NOT XP. Hence, the advice I got was to "switch off" this feature in Windows10 by adding the following registry entry in HKEY_LOCAL_MACHINE\SYSTEM\ControlSet001\Control\GraphicsDrivers: QWord "TdrLevel" with value "0". This, so to say, imitates what I had in WindowsXP, i.e. NO Tdr at all. However, the problem is that once I add this registry entry and restart the PC and Boinc, Boinc tells me "no usable GPU found" in the event log. So, it seems that BOINC is making a check regarding presence or absence of TDR - is this correct? If so, is there a way to eliminate this check? |
Copyright © 2025 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.