ATLAS Simulation v2.01 (vbox_64) windows_x86_64 failing after several hours with error 196 - EXIT_DISK_LIMIT_EXCEEDED


Advanced search

Questions and Answers : Windows : ATLAS Simulation v2.01 (vbox_64) windows_x86_64 failing after several hours with error 196 - EXIT_DISK_LIMIT_EXCEEDED

Author Message
spiko
Send message
Joined: 7 Jul 16
Posts: 2
Credit: 35,023
RAC: 0
    
Message 6126 - Posted: 8 Feb 2017, 17:53:20 UTC

Hi, I don't know how long this has been going on - I noticed just recently that I have several tasks with "computation error".

It happened near the end of work on a task (6-18 hours)

Error is claimed to be due to "Disk usage limit exceeded"

I spent an hour or so debugging the issue but it turned out to be very simple so I'd like to share it: Under disk limits in BOINC I had one option which I didn't notice, but was responsible for this error: Use at most "10%" of disk drive. This comes to about 6Gb and at the end of computation, the .vdi and other files were over that I presume.

I don't know how this could be detected before the size of task grew over the limit, but it sucks that so many hours of computing had to be discarded due to it.

PHILIPPE
Send message
Joined: 24 Jul 16
Posts: 84
Credit: 53,413
RAC: 0
    
Message 6127 - Posted: 8 Feb 2017, 21:19:08 UTC - in response to Message 6126.

Hi , Spiko,

this error seems to be present in various projects.

2 messages in your logs:

1° )

<core_client_version>7.6.33</core_client_version>
<![CDATA[
<message>
Disk usage limit exceeded
</message>
<stderr_txt>


2°)
09:19:15.938324 ERROR [COM]: aRC=VBOX_E_OBJECT_IN_USE (0x80bb000c) aIID={4afe423b-43e0-e9d0-82e8-ceb307940dda} aComponent={MediumWrap} aText={Medium 'C:\Program Files\Oracle\VirtualBox\VBoxGuestAdditions.iso' cannot be closed because it is still attached to 5 virtual machines}, preserve=false aResultDetail=0



the first error occurs when the vdi file in the directory slots is greater than 4GBytes.Boinc has difficulties to handle such file.So when the file grows up , it is not deleted and all the next tasks computed in this slot are corrupted.(message)

To solve the problem , you have to :
In the boinc manager
-click no more work
-upload all the results by clicking" update the project"(to avoid losing your credits)
-suspend all the idle tasks
-abort the tasks running
-Open virtualBox and delete all the failed VM (marked with a red sign).(right click on them)
If there is nothing , next step
-Go and see in your slots if there are remnants of old tasks not deleted (sort with the date column)(Delete files too old).
-Reboot your computer.
-Open Boinc manager and resume project.

If errors still present then un-attach and re-attach the project.
If no result ,change the virtualbox version used, in order to initialize the registry and delete the past links and reboot.

A faulty wu can produce errors and if you are unfortunate a set of faulty wus can be sent to your computer.This is sometimes perturbating and difficult to solve.
So keep trusting in yourself and in your computer...Long is the road...
These clean actions work in 90% of the usual cases.

spiko
Send message
Joined: 7 Jul 16
Posts: 2
Credit: 35,023
RAC: 0
    
Message 6128 - Posted: 8 Feb 2017, 23:06:30 UTC - in response to Message 6127.

Thank you for the reply. Actually I have no stale .vdi files in the slots directory over 4Gb. There are some vdi's over 3Gb, but they seem current, not remnants from an older task.

I did have another unit fail on me just now, accumulating 3gigs of save files + 3.2 gigs of .vdi so that might have been over the 6Gb limit I've seen mentioned somewhere.

For now I will just pause atlas and reenable it on BOINC update to see if it works.

Have fun.

Questions and Answers : Windows : ATLAS Simulation v2.01 (vbox_64) windows_x86_64 failing after several hours with error 196 - EXIT_DISK_LIMIT_EXCEEDED