| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: Computational Error, Linux 5.10.8, Discovering Dengue Drugs |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 20
|
|
| Author |
|
|
abennett
Cruncher Joined: Dec 11, 2005 Post Count: 10 Status: Offline Project Badges: 
|
Over the past weekend, I checked my Linux box and only one cruncher was running on my dual core machine. Checking the BOINC Manager, several tasks had "Computational Error" as the status.
I went through a process of stopping work, stopping network traffic, restarting, and eventually rebooting the machine. Eventually both cores were occupied with tasks. Monday night I came home and the machine was in the same condition again. Several Discovering Dengue Drugs had computational error. I stopped work, stopped network traffic, and forced a communications retry. Several tasks came down - 4 Dengue and 2 HPF. The BOINC manager showed that a Dengue task would start, then Computational Error, and it would flip to the next task. Once it got to a HPF task, it began crunching. I'm guessing that the server sending Dengue packets is sending malformed packets at the moment. Why can't the client handle these situations better? I'd rather not have to check my machine to make sure the clients are crunching. |
|||
|
||||
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Where did you get 5.10.8 from? On what Linux distro?
----------------------------------------WCG's recommended release is 5.10.21. Get it here: http://www.worldcommunitygrid.org/reg/ms/viewDownloadBoinc.do?newMember=false Many run Linux in duo or quad configurations without issue. If you provide us with a message log sequence showing from download to start and faulting of job, we can have a closer look.
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
abennett
Cruncher Joined: Dec 11, 2005 Post Count: 10 Status: Offline Project Badges:
|
5.10.8 is what is in Ubuntu's repositories. Doesn't WCG have a Ubuntu liason who gets the latest software drops into their repositories, such that users get updated during normal system updates? That'd be so nice!
So here is a recent log sequence showing the download of a job, the fault, and termination. Note the quick time sequences (seconds or less), DDD block dddt0201i0537_ZINC04736911-0000_00_0201, and it's checksum error. Tue 27 Nov 2007 12:06:45 PM PST|World Community Grid|[file_xfer] Started download of file dddt0201i0537_ZINC04736911-0000_00_ZINC04736911-0000.pdbqt Tue 27 Nov 2007 12:06:45 PM PST|World Community Grid|Starting dddt0201i0537_ZINC07822137-0000_00_0 Tue 27 Nov 2007 12:06:45 PM PST|World Community Grid|Starting task dddt0201i0537_ZINC07822137-0000_00_0 using dddt version 510 Tue 27 Nov 2007 12:06:46 PM PST|World Community Grid|[file_xfer] Finished download of file dddt0201i0537_ZINC04736911-0000_00_0201.pdbqt Tue 27 Nov 2007 12:06:46 PM PST|World Community Grid|[file_xfer] Throughput 68198 bytes/sec Tue 27 Nov 2007 12:06:46 PM PST|World Community Grid|[file_xfer] Started download of file dddt0201i0537_ZINC07155893-0000_00_ZINC07155893-0000.pdbqt Tue 27 Nov 2007 12:06:46 PM PST|World Community Grid|[error] Checksum or signature error for dddt0201i0537_ZINC04736911-0000_00_0201.pdbqt Tue 27 Nov 2007 12:06:46 PM PST|World Community Grid|Computation for task dddt0201i0537_ZINC07822137-0000_00_0 finished Tue 27 Nov 2007 12:06:46 PM PST|World Community Grid|Output file dddt0201i0537_ZINC07822137-0000_00_0_0 for task dddt0201i0537_ZINC07822137-0000_00_0 absent Tue 27 Nov 2007 12:06:46 PM PST|World Community Grid|Output file dddt0201i0537_ZINC07822137-0000_00_0_1 for task dddt0201i0537_ZINC07822137-0000_00_0 absent Tue 27 Nov 2007 12:06:47 PM PST|World Community Grid|[file_xfer] Finished download of file dddt0201i0537_ZINC07155893-0000_00_ZINC07155893-0000.pdbqt Tue 27 Nov 2007 12:06:47 PM PST|World Community Grid|[file_xfer] Throughput 1899 bytes/sec Tue 27 Nov 2007 12:06:47 PM PST|World Community Grid|[file_xfer] Started download of file dddt0201i0537_ZINC07155893-0000_00_AD4_parameters.dat Tue 27 Nov 2007 12:06:49 PM PST|World Community Grid|[file_xfer] Finished download of file dddt0201i0537_ZINC04736911-0000_00_ZINC04736911-0000.pdbqt |
||
|
|
Zoot Nerper
Cruncher Joined: Nov 21, 2004 Post Count: 19 Status: Offline Project Badges: 
|
Yes, I'd love someone to update the Ubuntu repositories with the latest version of Boinc. Pretty please
 (whether it's that simple I don't know) |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
I tried to follow the instructions on the setup up page that shows after downloading the WCG provided version:
----------------------------------------http://boincdl.ssl.berkeley.edu/dl/boinc_5.10.21_i686-pc-linux-gnu.sh http://www.worldcommunitygrid.org/reg/ms/viewThankYouAll.do?platformPick=linux Ubuntu makers are likely in direct interface with Berkeley. I dont see WCG interfacing with anyone compiling Linux distro's (seems that some Policy change at the developers was made to make the Linux version first line Ubuntu fit (?). WCG does tell Berkeley if something is fit or not.
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Nov 28, 2007 12:47:11 PM] |
||
|
|
abennett
Cruncher Joined: Dec 11, 2005 Post Count: 10 Status: Offline Project Badges:
|
*bump*
So forgetting my thoughts about BOINC and Ubuntu repositories, any thoughts on my theory that the servers are sending corrupted tasks to clients? |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I checked my DDD-t results, and there are no failed ones in the 6 pages recorded (oldest returned on the 19th). I am using Ubuntu 7.10 and whatever version of the client was stable on the BOINC site a few months ago.
|
||
|
|
abennett
Cruncher Joined: Dec 11, 2005 Post Count: 10 Status: Offline Project Badges:
|
So then what causes a "computational error" message in the client? Especially in my case, right after downloading a task (versus say chomping on a task for a few hours and then hitting a computational error).
|
||
|
|
twilyth
Master Cruncher US Joined: Mar 30, 2007 Post Count: 2130 Status: Offline Project Badges: 
|
So then what causes a "computational error" message in the client? Especially in my case, right after downloading a task (versus say chomping on a task for a few hours and then hitting a computational error). I've had the same problem with different versions of boinc including 5.10.8 on windows xp. I contacted support last week and I'm still waiting for them to look into it. The most they've been able to tell me so far is "check the error message" - not exactly helpful. I've since upgraded the one 5.10.8 machine to the 5.10.30 release but I don't think that is the problem. This issue has come up before according to a google search I did and it doesn't seem to have been resolved. Maybe in another week or two WCG support might actually get back to me on this. I'll let you know. One odd behavior I did notice is that just before all the cached wu's got tagged with a computational error, there was network access to isaac.ssl.berkley.edu. Don't know what the significance of that is.   |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
abennett, it is a checksum error. If there was a problem with the workunit, then the checksum would be fine.
A checksum error shows that something went wrong between downloading and starting the work unit. This usually means the problem is unique to you. twilyth, you say "the same problem". Are you sure? Can you post the relevant messages? |
||
|
|
|