LHC@home

This new version for Windows 64bit contains a number of improvements that have been evaluated on the development project. The main change is that the vboxwrapper uses the VBoxManage interface rather than the VBox API and hence aligns it with the Linux and Mac versions.

VM images that have been cleaned out from their respective slots are still leaving behind references in Vbox which have to be manually removed.

We are looking into this as it is something that has been happening with previous versions too.

THANKS for updating the VB start time to 20 minutes here too.

(yes I am the DSL trouble maker from over at vLHC-dev )

---

Volunteer Mad Scientist For Life

I haven't provided an analysis for some time so here is the result of investigating errors with this version. Of all the tasks that were run over the past 24 hours, 25% of machines are failing. Here is a breakdown of the exit status codes and the percentage of machines that are failing.

• 8.33% 194
  
• 7.84% -1073741819
  
• 6.62% 206
  
• 3.68% 1
  
• 2.94% -152
  
• 1.96% -203
  
• 1.72% 203
  
• 0.98% -2147467259
  
• 0.74% -2135228415
  
• 0.74% -226
  
• 0.74% -108
  
• 0.49% 255
  
• 0.25% -2135228412
  
• 0.25% -186
  
• 0.25% -185
  

Please check your tasks. A project reset may fix some of these errors, so try that first. Feel free to ask any questions on errors here and I will try answer. It will help if you post the URL to the task that is failing.

So far I have 102 Valids and the 23 errors are just that slow dsl problem I have so those tasks most likely are ones that tried to start after noon.

This morning it is slow as a snail up a sequoia tree.....and again I am tempted to call about switching to satellite again.

29 valid recently, no failures and 1 Abort (to allow space to run a few, very short, Sixtracks), although I did have to manually remove 12 yellow triangles from Vbox.

I have 373 good and one bad, so very reilable to me.

2017-06-04 21:57:57 (11124): Error in start VM for VM: -2135228415
Command:
VBoxManage -q startvm "boinc_24471e3e34f08fed" --type headless
Output:
VBoxManage.exe: error: Could not find a registered machine named 'boinc_24471e3e34f08fed'

It is very important to implement a check, if a job has a chance to finish within the 18h Limit.

This has been an issue for a long time and has not been addressed.

IF THE JOB HITS THE 18h MARK, it will be terminated and no results uploaded.

The amount of wasted computing time is considerable-especially as the unfinished job will be resend to (most likely) fail on an other machine.

After 12 hours no more jobs are started hence jobs are given a 6 hour grace period to exit. If it doesn't finish within that time then there is a problem with the job. If the VM is terminated at 18 hours, the task is reported successful.

the task is reported successful

That may be in boinc terms.

However, i had jobs, that appeared to be running properly, but where just running out of time, therefore not finishing.

Or, there was the first job in a boinc Task, running for 18h and being terminated.

This should be avoided.