| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: Computation Error with some projects |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 19
|
|
| Author |
|
|
Tom Moorman
Cruncher Joined: Mar 3, 2007 Post Count: 4 Status: Offline Project Badges: 
|
I have been running WCG on BOINC for several years now with no problems.
A couple of days ago I started getting "computation error" messages for every work unit in the Mapping Cancer Markers, Smash Childhood Cancer and FightAIDS@Home projects. To test whether it was BOINC I resumed work on SETI and all work units, both cpu and gpu, work without issue. I went back to WCG and tried a new project (to me): Microbiome Immunity Project. In this project I have successfully completed several work units. I tried Smash Childhood Cancer again and got the same computation error. The event log shows this for the work unit: Thu 15 Feb 2018 12:07:17 AM EST | World Community Grid | task SCC1_0001750_Lin-CSD-A_21950_0 resumed by user Thu 15 Feb 2018 12:07:17 AM EST | World Community Grid | Starting task SCC1_0001750_Lin-CSD-A_21950_0 Thu 15 Feb 2018 12:07:18 AM EST | World Community Grid | Computation for task SCC1_0001750_Lin-CSD-A_21950_0 finished Thu 15 Feb 2018 12:07:18 AM EST | World Community Grid | Output file SCC1_0001750_Lin-CSD-A_21950_0_r1136150297_0 for task SCC1_0001750_Lin-CSD-A_21950_0 absent I am running Arch Linux (64 bit, fully updated). Is this me or somethiing else? |
||
|
|
pcwr
Ace Cruncher England Joined: Sep 17, 2005 Post Count: 10903 Status: Offline Project Badges: 
|
When was the last time you did a reboot?
----------------------------------------Which version of BOINC are you running? Patrick  |
||
|
|
Tom Moorman
Cruncher Joined: Mar 3, 2007 Post Count: 4 Status: Offline Project Badges:
|
Rebooted in the last 24 hours.
I am running 7.8.4 of BOINC Manager from the Arch Linux repository package called boinc-7.8.4-1-x86_64.pkg.tar.xz |
||
|
|
Tom Moorman
Cruncher Joined: Mar 3, 2007 Post Count: 4 Status: Offline Project Badges:
|
As per pacman:
Name : boinc Version : 7.8.4-1 Description : Berkeley Open Infrastructure for Network Computing for desktop Architecture : x86_64 URL : http://boinc.berkeley.edu/ Licenses : LGPL Groups : None Provides : None Depends On : libxss libnotify wxgtk3 webkit2gtk curl sqlite3 Optional Deps : None Required By : None Optional For : None Conflicts With : None Replaces : None Installed Size : 8.69 MiB Packager : Felix Yan <felixonmars@archlinux.org> Build Date : Fri 24 Nov 2017 10:03:10 AM EST Install Date : Wed 07 Feb 2018 10:58:39 AM EST Install Reason : Explicitly installed Install Script : Yes Validated By : Signature Dependencies: libxss: 1.2.2-2 libnotify: 0.7.7-1 wxgtk3: 3.0.3.1-11 webkit2gtk: 2.18.6-1 curl: 7.58.0-1 sqlite: 3.22.0-1 (Provides sqlite3=3.22.0) I am running Linux kernel 4.15.3-1 |
||
|
|
redmaw
Cruncher Joined: Apr 14, 2010 Post Count: 6 Status: Offline Project Badges: 
|
I think I am seeing the same thing. After doing a system update and restarting every project started failing with computation errors. The errors I checked all reported segmentation faults as the reason.
----------------------------------------A small percentage of tasks are not failing right away though so for now I am letting those run to see if the can complete without error. Was this issue ever resolved for you? [Edit 1 times, last edit by redmaw at Feb 25, 2018 1:44:46 AM] |
||
|
|
[AF>Libristes]Maeda
Cruncher Joined: Sep 1, 2011 Post Count: 43 Status: Offline Project Badges: 
|
Same here, just updated my system
----------------------------------------In fact, it's since the 13th of February at least. I just stop asking new WU so that I stop sending WU in error.  My other computer with Debian Stretch don't any errors. [Edit 1 times, last edit by [AF>Libristes]Maeda at Feb 25, 2018 9:36:06 PM] |
|||
|
||||
|
KerSamson
Master Cruncher Switzerland Joined: Jan 29, 2007 Post Count: 1684 Status: Offline Project Badges: 
|
Since the inappropriate patches delivered by Intel because of Meltdown and Spectre 1 & 2 early January, I reduced drastically the system updates of my Linux systems, carefully waiting for a better time.
----------------------------------------A couple of days ago, new patches have been announced. I don't know if there is a relationship between these new patches and the reported segmentation fault issues. Because of the nature of the patches, I would not ignore this eventuality. Cheers, Yves --- PS: If possible, you should maybe try at first to uninstall the Intel microcode update (see proprietary drivers). |
||
|
|
[AF>Libristes]Maeda
Cruncher Joined: Sep 1, 2011 Post Count: 43 Status: Offline Project Badges:
|
If I do :
grep . /sys/devices/system/cpu/vulnerabilities/* And grep microcode /proc/cpuinfo |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Intel have replaced some firmware updates and have re-instated some withdrawn ones. They seem to be close to being in 'headless chicken' mode -- I'm going to wait a while before I update mine.
----------------------------------------The latest press release appears to be this one: https://newsroom.intel.com/news/latest-intel-...dated-firmware-available/ It provides links to information that may help you. [Edit 1 times, last edit by Former Member at Feb 26, 2018 12:37:05 AM] |
||
|
|
redmaw
Cruncher Joined: Apr 14, 2010 Post Count: 6 Status: Offline Project Badges:
|
Waiting on updating the kernel and microcode (if you even do) is probably a good idea, however my update only pulled updated packages so I am fairly confident that is not the issue.
After some poking about the BOINC documentation I suspect the update broke or changed something the applications depend on as some of them still work while others segfault even when invoked from the command line with no arguments. The problem for me is not occurring during actual run time but immediately when boinc executes the application. I also tried the latest boinc client and manager but nothing changed. It doesn't look like the source for the applications is available though so I can't think of anyway to debug them. |
||
|
|
|